llama 7B

1431 records across 166 hardware.

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
H200llama 7B Q4_0llama.cpp (CUDA)337.611633.96––tg1282025-11-15community–source
H100llama 7B Q4_0llama.cpp (CUDA)328.711564.03––tg1282025-11-15community–source
B200llama 7B Q4_0llama.cpp (CUDA)315.8511510.56––tg1282025-10-25community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)315.317107.92––tg1282025-12-04community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)314.014805.32––tg1282025-12-04community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)312.3316070.88––tg1282025-11-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)310.916508.13––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)308.886695.94––tg1282026-03-18community–source
H200llama 7B Q4_0llama.cpp (CUDA)307.7910428.65––tg1282025-11-15community–source
H100llama 7B Q4_0llama.cpp (CUDA)303.0810162.9––tg1282025-11-15community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)301.9215505.79––tg1282025-12-09community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)300.414970.15––tg1282025-10-23community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)299.9214036.03––tg1282025-11-07community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)299.8514013.74––tg1282025-11-07community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)298.56695.4––tg1282026-03-20community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)298.1813951.14––tg1282025-11-07community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)297.9814131.49––tg1282025-12-04community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)297.8714988.21––tg1282025-12-04community–source
B200llama 7B Q4_0llama.cpp (CUDA)297.6810531.52––tg1282025-10-25community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)294.9513964.52––tg1282025-11-08community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)290.0214073.41––tg1282025-10-23community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)289.314463.16––tg1282026-06-11community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)287.5513451.73––tg1282025-12-09community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)287.2913403.7––tg1282026-05-08community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)286.5116195.14––tg1282025-09-26community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)285.5115602.75––tg1282025-11-07community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)284.0316053.65––tg1282025-11-07community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)283.6915915.61––tg1282025-11-07community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)281.1116618.98––tg1282025-09-05community–source
H100llama 7B Q4_0llama.cpp (CUDA)280.7411263.29––tg1282025-09-05community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)280.2813143.45––tg1282025-11-07community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)278.226287.12––tg1282026-03-20community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)277.2114142.08––tg1282025-09-26community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)275.7918012.64––tg1282026-04-02community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)275.7616652.18––tg1282025-08-11community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)274.214854.63––tg1282025-09-05community–source
RTX 5090llama 7B Q4_0llama.cpp (Vulkan)273.6811796.38––tg1282025-10-11community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)271.2612242.46––tg1282026-05-08community–source
RTX PRO 6000llama 7B Q4_0llama.cpp (CUDA)270.011443.35––tg1282026-01-22community–source
RTX PRO 6000llama 7B Q4_0llama.cpp (CUDA)269.9612231.9––tg1282026-01-22community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)268.5214401.14––tg1282025-08-11community–source
H100llama 7B Q4_0llama.cpp (CUDA)267.819918.34––tg1282025-09-05community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)266.715094.35––tg1282025-11-07community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)264.6615928.07––tg1282026-04-02community–source
RTX 5090llama 7B Q4_0llama.cpp (Vulkan)263.6310381.64––tg1282025-10-11community–source
RTX Pro 6000 Blackwellllama 7B Q4_0llama.cpp (CUDA)260.4212742.48––tg1282026-06-11community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)248.5716041.54––tg1282025-08-02community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)239.6214751.98––tg1282025-08-02community–source
H100llama 7B Q4_0llama.cpp (CUDA)229.858457.52––tg1282025-09-05communitycontradictionsource
RTX Pro 5000 Blackwellllama 7B Q4_0llama.cpp (CUDA)226.3111046.55––tg1282026-02-28community–source
H100llama 7B Q4_0llama.cpp (CUDA)219.237350.98––tg1282025-09-05communitycontradictionsource
A100llama 7B Q4_0llama.cpp (CUDA)215.115422.2––tg1282025-11-15community–source
RTX Pro 5000 Blackwellllama 7B Q4_0llama.cpp (CUDA)213.549952.91––tg1282026-02-28community–source
A100llama 7B Q4_0llama.cpp (CUDA)200.95285.96––tg1282025-09-05community–source
A100llama 7B Q4_0llama.cpp (CUDA)200.054967.14––tg1282025-11-15community–source
A100llama 7B Q4_0llama.cpp (CUDA)196.265295.94––tg1282025-09-05community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)195.33332.9––tg1282025-11-05community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)192.178054.59––tg1282026-01-31community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)192.178054.59––tg1282026-01-31community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)191.5712506.97––tg1282025-08-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)191.283531.93––tg1282025-11-05community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)190.923889.54––tg1282026-09-10community–source
A100llama 7B Q4_0llama.cpp (CUDA)190.884849.53––tg1282025-09-05community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)190.787022.52––tg1282026-01-24community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)190.787022.52––tg1282026-01-24community–source
RTX 4090llama 7B Q4_0llama.cpp (Vulkan)190.110830.41––tg1282025-09-24community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)189.3412415.31––tg1282025-08-24community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)189.3310293.86––tg1282025-08-24community–source
RTX 4090llama 7B Q4_0llama.cpp (CUDA)189.0714753.2––tg1282025-08-03community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)188.9614770.63––tg1282025-08-06community–source
RTX 4090llama 7B Q4_0llama.cpp (Vulkan)187.979452.03––tg1282025-09-24community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)187.8310259.68––tg1282025-08-24community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)186.2111992.7––tg1282025-08-06community–source
A100llama 7B Q4_0llama.cpp (CUDA)185.914883.69––tg1282025-09-05community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)185.17444.99––tg1282026-01-31community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)185.17444.99––tg1282026-01-31community–source
RTX 4090llama 7B Q4_0llama.cpp (CUDA)185.0414472.84––tg1282025-07-08community–source
GeForce RTX 5080llama 7B Q4_0llama.cpp (CUDA)184.689487.7––tg1282025-08-28community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)183.16840.43––tg1282026-01-24community–source
RTX 5080llama 7B Q4_0llama.cpp (Vulkan)183.16840.43––tg1282026-01-24community–source
GeForce RTX 5080llama 7B Q4_0llama.cpp (CUDA)183.049644.43––tg1282025-12-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)182.633726.99––tg1282026-09-10community–source
GeForce RTX 5070 Tillama 7B Q4_0llama.cpp (CUDA)182.438419.56––tg1282025-12-07community–source
GeForce RTX 5080llama 7B Q4_0llama.cpp (CUDA)181.998297.36––tg1282025-08-28community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)179.663932.57––tg1282026-08-23communitycontradictionsource
RTX 4090llama 7B Q4_0llama.cpp (Vulkan)179.629481.99––tg1282025-08-03community–source
NVIDIA RTX 6000llama 7B Q4_0llama.cpp (CUDA)179.4710576.85––tg1282025-09-22community–source
M5 Ultrallama 7B 6.74BQ4_0llama.cpp (Metal)179.14944.66––tg1282026-09-30community–source
GeForce RTX 5080llama 7B Q4_0llama.cpp (CUDA)177.997405.11––tg1282025-12-16community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)177.795992.1––tg1282026-05-03community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)177.755973.72––tg1282026-09-07community–source
A100llama 7B Q4_0llama.cpp (CUDA)176.885141.45––tg1282025-05-22community–source
GeForce RTX 5070 Tillama 7B Q4_0llama.cpp (CUDA)176.856952.38––tg1282025-12-07community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)176.187567.94––tg1282026-04-24community–source
NVIDIA RTX 6000llama 7B Q4_0llama.cpp (CUDA)176.079229.23––tg1282025-09-22community–source
RTX PRO 4500 Blackwellllama 7B Q4_0llama.cpp (CUDA)175.578149.86––tg1282025-11-14community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)175.485874.09––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)173.813153.9––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)173.733259.34––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)173.363253.68––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)173.133250.86––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)172.863290.92––tg1282026-03-13community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)172.842706.82––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)172.792806.12––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)172.762801.46––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)172.712805.99––tg1282025-12-11community–source
GeForce RTX 3090 Tillama 7B Q4_0llama.cpp (CUDA)172.266924.01––tg1282025-08-01community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)171.615200.68––tg1282026-05-03community–source
GeForce RTX 3090 Tillama 7B Q4_0llama.cpp (CUDA)171.196567.49––tg1282025-08-01community–source
RTX 4090llama 7B Q4_0llama.cpp (Vulkan)170.958534.56––tg1282025-07-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)170.693692.87––tg1282026-08-23community–source
M4llama 7B Q4_0llama.cpp (Vulkan)170.567064.4––tg1282025-12-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)170.443374.48––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)170.133178.79––tg1282025-11-20community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)170.1214017.58––tg1282025-08-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)170.123874.25––tg1282025-11-05community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)169.7412416.64––tg1282025-08-27community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)169.7412416.64––tg1282025-08-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)169.382994.51––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)169.063138.42––tg1282025-12-11community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)168.955153.91––tg1282026-09-07community–source
RTX PRO 4500 Blackwellllama 7B Q4_0llama.cpp (CUDA)168.97251.66––tg1282025-11-14community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)168.835349.43––tg1282026-05-03community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.773121.91––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.742622.48––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.633128.43––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.623479.41––tg1282026-03-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.532333.75––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.532333.75––tg1282025-11-20community–source
A100llama 7B Q4_0llama.cpp (CUDA)168.234865.98––tg1282025-05-22community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (CUDA)168.27802.37––tg1282025-05-10community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)168.042714.98––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)167.792716.65––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)167.782727.7––tg1282025-12-11community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)167.5310712.64––tg1282025-08-27community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)167.5310712.64––tg1282025-08-27community–source
GeForce RTX 4090llama 7B Q4_0llama.cpp (CUDA)167.5211773.24––tg1282025-08-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)167.423042.23––tg1282025-11-20community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)167.395134.61––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)167.312175.84––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)167.312175.84––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)167.113552.27––tg1282025-11-05community–source
RTX 4090llama 7B Q4_0llama.cpp (Vulkan)166.459032.44––tg1282025-07-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)166.252044.99––tg1282025-10-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.943143.31––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.573170.06––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.383170.2––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.382148.72––tg1282025-11-15community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.233165.77––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)165.082281.62––tg1282025-10-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)164.842750.74––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)164.322805.52––tg1282025-12-11community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)164.317603.2––tg1282026-09-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)164.283181.81––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)164.222814.04––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)164.212807.29––tg1282025-12-11community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)164.054666.15––tg1282026-05-03community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)163.774403.4––tg1282026-09-10community–source
RTX 3090llama 7B Q4_0llama.cpp (CUDA)163.695781.6––tg1282025-09-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)163.01997.19––tg1282025-10-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)162.782284.71––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)162.782284.71––tg1282025-11-20community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)162.284732.33––tg1282025-09-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)162.162254.0––tg1282025-10-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)162.022170.71––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)162.022170.71––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.633044.63––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.533164.42––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.482973.55––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.463453.79––tg1282026-02-16communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.393069.84––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)161.073338.82––tg1282026-03-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)160.783051.47––tg1282025-12-11community–source
M4llama 7B Q4_0llama.cpp (Vulkan)160.786389.86––tg1282025-12-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)160.773056.15––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)160.772653.13––tg1282025-12-11community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)160.675051.48––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)160.22739.28––tg1282025-12-11community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)160.134298.97––tg1282025-09-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)159.862727.9––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)159.782726.36––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)159.462162.71––tg1282025-11-15community–source
RTX 3090llama 7B Q4_0llama.cpp (CUDA)159.365100.68––tg1282025-09-01community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)158.147641.89––tg1282026-05-26community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)157.913266.07––tg1282025-02-06community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)157.846663.79150.0–tg1282026-09-16community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)157.824409.02––tg1282025-09-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)157.413422.83––tg1282026-08-23community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)156.574892.33––tg1282025-09-01community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)156.496667.06––tg1282026-05-26community–source
A100llama 7B Q4_0llama.cpp (CUDA)155.723016.29––tg1282026-07-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)155.694145.55––tg1282026-09-10community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)155.365932.44––tg1282025-08-02community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)154.323292.37––tg1282026-02-16community–source
RTX 5090llama 7B Q4_0llama.cpp (Vulkan)154.316011.18160.0–tg1282026-09-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)153.913434.01––tg1282025-08-06community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)153.884190.07––tg1282026-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)153.723599.57––tg1282025-05-26community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (CUDA)153.557623.55––tg1282025-05-23community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)153.014425.82––tg1282025-12-11community–source
V100llama 7B Q4_0llama.cpp (CUDA)152.363134.81––tg1282025-11-19community–source
V100llama 7B Q4_0llama.cpp (CUDA)152.333537.71––tg1282025-12-17community–source
V100llama 7B Q4_0llama.cpp (CUDA)151.883133.71––tg1282025-11-19community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)151.654356.94––tg1282025-12-11community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)151.654435.33––tg1282025-12-11community–source
GeForce RTX 3090llama 7B Q4_0llama.cpp (CUDA)151.045405.83––tg1282025-08-02community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)150.973229.07––tg1282026-03-13communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)150.863529.1––tg1282025-07-22community–source
V100llama 7B Q4_0llama.cpp (CUDA)150.683075.88––tg1282025-11-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)150.43458.31––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)150.342369.54––tg1282025-11-15communitycontradictionsource
RTX 4080 SUPERllama 7B Q4_0llama.cpp (Vulkan)150.28007.37––tg1282025-10-18community–source
RTX 5090llama 7B Q4_0llama.cpp (Vulkan)149.415546.39––tg1282026-09-17community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (CUDA)149.286870.08––tg1282025-05-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)149.043252.09––tg1282026-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)148.343134.2––tg1282025-08-06community–source
GeForce RTX 4080 Superllama 7B Q4_0llama.cpp (CUDA)148.338125.15––tg1282025-10-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)148.213236.63––tg1282025-02-06community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)148.013288.72––tg1282025-08-06community–source
GeForce RTX 4080 Superllama 7B Q4_0llama.cpp (CUDA)147.489439.01––tg1282025-10-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)147.251943.62––tg1282025-06-23community–source
RTX 4080 SUPERllama 7B Q4_0llama.cpp (Vulkan)147.137101.18––tg1282025-10-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)147.063875.83––tg1282026-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)146.661986.43––tg1282025-05-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)145.781952.7––tg1282025-06-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)145.73225.19––tg1282025-07-22community–source
V100llama 7B Q4_0llama.cpp (CUDA)145.453434.83––tg1282025-12-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)145.351937.34––tg1282025-07-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)145.323966.95––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)145.233633.86––tg1282025-08-06community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)145.03489.67––tg1282025-07-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)144.93419.13––tg1282025-05-26community–source
A6000llama 7B Q4_0llama.cpp (CUDA)144.875662.39––tg1282025-09-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)144.843552.47––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)143.992062.17––tg1282025-01-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)143.983743.15––tg1282025-05-26community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)143.885438.27––tg1282026-04-03community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)143.752222.31––tg1282025-11-15communitycontradictionsource
GeForce RTX 4080llama 7B Q4_0llama.cpp (CUDA)143.479205.93––tg1282025-08-07community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)143.463877.12––tg1282025-12-11community–source
RTX 3080llama 7B Q4_0llama.cpp (Vulkan)143.444889.13––tg1282026-04-03community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)142.886004.06––tg1282026-06-05community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)142.764033.29––tg1282025-12-11community–source
GeForce RTX 4080llama 7B Q4_0llama.cpp (CUDA)142.498031.64––tg1282025-08-07community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)142.173764.9––tg1282026-08-23community–source
M2llama 7B Q4_0llama.cpp (Vulkan)142.131411.25––tg1282025-11-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)141.523187.61––tg1282025-06-13community–source
A100llama 7B Q4_0llama.cpp (CUDA)141.322698.95––tg1282026-07-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.963598.42––tg1282025-07-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.943516.41––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.943516.41––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.833577.58––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)140.762001.05––tg1282025-05-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)140.682039.24––tg1282025-01-08community–source
A5000llama 7B Q4_0llama.cpp (Vulkan)140.434071.22––tg1282025-09-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.273588.15––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)140.273588.15––tg1282025-11-20community–source
V100llama 7B Q4_0llama.cpp (CUDA)140.233261.46––tg1282025-12-17community–source
V100llama 7B Q4_0llama.cpp (CUDA)140.03235.72––tg1282025-11-19community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)139.955569.56––tg1282025-08-01community–source
A5000llama 7B Q4_0llama.cpp (Vulkan)139.893641.55––tg1282025-09-24community–source
V100llama 7B Q4_0llama.cpp (CUDA)139.893235.01––tg1282025-11-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)139.893623.53––tg1282026-03-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)139.813321.11––tg1282026-03-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)139.683482.15––tg1282025-11-20community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)139.655013.86––tg1282025-08-01community–source
V100llama 7B Q4_0llama.cpp (CUDA)139.113149.71––tg1282025-11-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)138.913782.88––tg1282025-12-10community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)138.843895.62––tg1282025-09-23community–source
A6000llama 7B Q4_0llama.cpp (CUDA)138.734913.93––tg1282025-09-01community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)138.453525.21––tg1282025-09-23community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)137.781619.74––tg1282025-08-23community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)137.314262.85––tg1282025-11-29community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)137.115036.04––tg1282026-01-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)136.981997.04––tg1282025-01-08community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)136.814543.96––tg1282025-07-08community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)136.645673.65––tg1282026-06-05community–source
RTX PRO 4000 Blackwellllama 7B Q4_0llama.cpp (CUDA)136.385674.44––tg1282025-11-21community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)136.242022.59––tg1282025-01-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)136.053537.94––tg1282026-08-23community–source
A5000llama 7B Q4_0llama.cpp (CUDA)135.834552.15––tg1282025-08-19community–source
RTX 4070 Ti SUPERllama 7B Q4_0llama.cpp (Vulkan)135.816801.18––tg1282025-09-24community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)135.651619.49––tg1282025-08-23community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (Vulkan)135.636213.63––tg1282025-05-23community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)135.074901.77––tg1282025-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)134.871151.72––tg1282026-03-01communitycontradictionsource
V100llama 7B Q4_0llama.cpp (CUDA)134.762973.78––tg1282025-09-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)134.461155.75––tg1282026-02-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)134.463245.11––tg1282026-03-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)134.073309.54––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)134.073309.54––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)134.013364.06––tg1282025-11-20community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (Vulkan)133.946614.86––tg1282025-05-23community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)133.634946.47––tg1282025-07-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)133.593365.22––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)133.593365.22––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)133.423110.71––tg1282025-06-23community–source
V100llama 7B Q4_0llama.cpp (CUDA)133.063160.43––tg1282025-12-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)133.021876.28––tg1282025-07-17community–source
GeForce RTX 4070 Ti Superllama 7B Q4_0llama.cpp (CUDA)132.857612.32––tg1282025-08-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)132.783278.06––tg1282025-11-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)132.753085.75––tg1282025-06-28community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)132.753316.94––tg1282026-08-23community–source
Radeon 880Mllama 7B Q4_0llama.cpp (ROCm)132.632727.83––tg1282025-10-28community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)132.273465.31––tg1282025-12-18community–source
GeForce RTX 4070 Ti Superllama 7B Q4_0llama.cpp (CUDA)132.266924.53––tg1282025-08-01community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)132.184293.66––tg1282025-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)132.173473.24––tg1282025-08-13community–source
Radeon 880Mllama 7B Q4_0llama.cpp (ROCm)132.122632.74––tg1282025-10-28community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)131.994273.68––tg1282025-11-29community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)131.953224.31––tg1282025-07-17community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)131.73541.67––tg1282025-12-10community–source
RTX PRO 4000 Blackwellllama 7B Q4_0llama.cpp (CUDA)131.664992.83––tg1282025-11-21community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)131.545048.07––tg1282026-01-24community–source
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (Vulkan)131.195411.28––tg1282026-09-12community–source
RTX 3080llama 7B Q4_0llama.cpp (CUDA)131.014499.47––tg1282024-12-18community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)131.015505.31––tg1282026-03-27community–source
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (Vulkan)130.974235.87––tg1282026-09-12community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)130.92925.56––tg1282025-06-23community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)130.573831.64––tg1282025-08-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)130.563461.64––tg1282025-08-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)130.121964.92––tg1282025-08-24community–source
A5000llama 7B Q4_0llama.cpp (CUDA)130.074028.16––tg1282025-08-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.82546.13––tg1282026-03-13communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.691313.1––tg1282026-03-01communitycontradictionsource
M2llama 7B Q4_0llama.cpp (Vulkan)129.581391.39––tg1282025-11-19community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.482281.29––tg1282025-12-11community–source
RTX 4070 Ti SUPERllama 7B Q4_0llama.cpp (Vulkan)129.456099.18––tg1282025-09-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)129.331253.3––tg1282026-02-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.32285.41––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.222561.41––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.21878.23––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.152278.5––tg1282025-12-11community–source
V100llama 7B Q4_0llama.cpp (CUDA)129.083042.64––tg1282025-09-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.072130.63––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)129.072590.43––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)128.952580.6––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)128.942581.45––tg1282025-12-11community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)128.82055.15––tg1282025-08-24community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)128.792871.79––tg1282025-06-28community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)128.534046.36––tg1282025-10-11community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)128.493586.22––tg1282025-09-23community–source
A100llama 7B Q4_0llama.cpp (CUDA)128.411175.43––tg1282026-05-26community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)128.385541.12––tg1282026-06-05community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)128.253386.75––tg1282025-08-08community–source
GeForce RTX 5070llama 7B Q4_0llama.cpp (CUDA)128.215783.44––tg1282025-08-20community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)127.693633.48––tg1282025-09-23community–source
GeForce RTX 5070llama 7B Q4_0llama.cpp (CUDA)127.545184.75––tg1282025-08-20community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)127.433698.73––tg1282025-08-13community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)127.384124.63––tg1282025-10-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)127.323486.59––tg1282025-07-17community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)127.265380.18––tg1282026-04-09community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)126.813493.01––tg1282025-12-18community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (Vulkan)126.675164.6––tg1282025-05-10community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)126.283730.42––tg1282025-12-10community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)126.063384.6––tg1282025-06-18community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)126.061888.77––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)126.061888.77––tg1282025-11-20communitycontradictionsource
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (Vulkan)125.965033.0––tg1282026-09-12community–source
RTX 5070 Tillama 7B Q4_0llama.cpp (Vulkan)125.896801.97––tg1282025-05-12community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)125.843243.15––tg1282025-08-02community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)125.644945.07––tg1282026-03-27community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)125.63336.74––tg1282025-06-18community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)125.453530.68––tg1282025-08-07community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)124.862197.05––tg1282026-01-29community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)124.753787.28––tg1282025-06-29community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)124.613674.25––tg1282025-08-08community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)124.333558.71––tg1282025-08-07community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)123.89994.51––tg1282025-09-27community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)123.723301.47––tg1282025-01-08community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)123.72307.03––tg1282025-06-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)123.71941.47––tg1282025-11-15communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)123.693405.36––tg1282025-07-27community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)123.182941.58––tg1282025-06-04community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)122.92530.61––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)122.863219.16––tg1282025-06-04community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)122.713557.68––tg1282025-08-02community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)122.614241.77––tg1282025-04-12communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)122.563169.7––tg1282025-06-28community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)122.543399.14––tg1282025-06-29community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)122.521002.13––tg1282025-09-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)122.063369.65––tg1282025-08-02community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)122.022272.76––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)121.873236.05––tg1282025-06-23community–source
A100llama 7B Q4_0llama.cpp (Vulkan)121.833103.32––tg1282025-05-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.622273.79––tg1282025-12-11community–source
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (ROCm)121.625039.78––tg1282026-09-12community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.612524.02––tg1282025-12-11community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)121.581638.62––tg1282025-09-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.572078.19––tg1282025-12-11communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.562276.59––tg1282025-12-11community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.551869.34––tg1282025-12-11communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.52512.95––tg1282025-12-11community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)121.41595.43––tg1282025-09-27community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)121.381998.36––tg1282025-10-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)121.332521.5––tg1282025-12-11community–source
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (ROCm)121.292755.01––tg1282026-09-12community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)121.252313.09––tg1282025-06-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)121.143401.14––tg1282025-07-27community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)121.044306.74––tg1282026-04-09community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)120.883379.59––tg1282025-07-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)120.733638.45––tg1282025-07-27community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)120.622701.13––tg1282025-06-04community–source
A100llama 7B Q4_0llama.cpp (Vulkan)120.533164.55––tg1282025-05-22community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)120.521707.58––tg1282025-06-29community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)120.462296.83––tg1282025-05-26community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)120.444516.92––tg1282025-04-12communitycontradictionsource
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)120.251573.21––tg1282025-09-27community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)120.121659.94––tg1282025-09-27community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)119.981873.97––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)119.981873.97––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)119.973408.06––tg1282025-11-20communitycontradictionsource
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)119.974824.33––tg1282026-03-25community–source
M5 Max (40 GPU)llama 7B Q4_0llama.cpp (Metal)119.923219.99–––2023-11-22community–source
M5 Maxllama 7B Q4_0llama.cpp (Metal)119.923219.99––tg1282026-08-25community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)119.891919.21––tg1282025-10-08community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (Vulkan)119.713164.1––tg1282025-08-15community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)119.691892.58––tg1282025-11-15communitycontradictionsource
Radeon RX 9070llama 7B Q4_0llama.cpp (Vulkan)119.512859.98––tg1282025-08-15community–source
M5 Maxllama 7B Q4_0llama.cpp (Metal)119.083347.44––tg1282026-06-22community–source
M5 Ultrallama 7B 6.74BQ8_0llama.cpp (Metal)119.014910.55––tg1282026-09-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)119.03429.68––tg1282025-12-10community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)118.713573.3––tg1282025-08-02community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)118.54368.74––tg1282025-01-16community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)118.272017.33––tg1282026-01-29community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)118.233323.43––tg1282025-06-28community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)117.722336.07––tg1282025-05-26community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)117.652609.51––tg1282025-07-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)117.613409.8––tg1282025-06-23community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)117.572076.96––tg1282025-08-23community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)117.452823.35––tg1282025-07-30community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)117.222206.24––tg1282025-03-18community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)117.165007.9––tg1282026-05-26communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)117.123605.23––tg1282025-07-27community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)117.113267.35––tg1282025-06-04community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)116.972053.11––tg1282025-09-27community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)116.954733.95––tg1282026-06-05community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)116.941662.15––tg1282025-08-23community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)116.833447.07––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)116.833447.07––tg1282025-11-20communitycontradictionsource
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (Vulkan)116.821689.15––tg1282025-06-29community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)116.72711.43––tg1282026-03-18community–source
Arc Pro B70llama 7B Q4_0llama.cpp (Vulkan)116.674651.26––tg1282026-08-26community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)116.611501.02––tg1282025-09-27community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)116.55167.28––tg1282026-03-25community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)116.465166.48––tg1282026-04-20community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)116.153098.38––tg1282025-07-30community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)116.112336.31––tg1282025-12-18community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)116.02967.12––tg1282025-08-01community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)115.844074.98––tg1282025-10-23community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)115.711967.78––tg1282025-09-27community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)115.55881.05––tg1282025-09-27community–source
M3 Ultrallama 7B Q4_0llama.cpp (Vulkan)115.541116.83––tg1282025-04-27community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)115.322452.68––tg1282025-10-23community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)115.113901.21––tg1282025-10-23community–source
M3 Ultrallama 7B Q4_0llama.cpp (Vulkan)114.551116.48––tg1282025-04-27community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)114.482381.77––tg1282025-10-23community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)114.181258.58––tg1282025-05-14community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)114.121432.7––tg1282025-05-14community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)114.023190.19––tg1282025-11-20communitycontradictionsource
M1 Ultrallama 7B Q4_0llama.cpp (Metal)113.791179.89––tg1282026-08-28community–source
RTX 4070 Tillama 7B Q4_0llama.cpp (Vulkan)113.525515.26––tg1282026-01-17community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)113.291003.78––tg1282025-09-27community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)112.753163.24––tg1282025-08-01community–source
Radeon AI PRO R9700llama 7B Q4_0llama.cpp (ROCm)112.724665.08––tg1282026-09-12community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)112.591632.95––tg1282025-10-15community–source
Arc Pro B70llama 7B Q4_0llama.cpp (Vulkan)112.434271.32––tg1282026-08-26community–source
GeForce RTX 5090llama 7B Q4_0llama.cpp (CUDA)112.365056.95––tg1282026-05-26communitycontradictionsource
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (ROCm)112.33261.75––tg1282025-07-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)112.182817.81––tg1282025-09-17community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)112.081434.13––tg1282026-08-26community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)112.011244.85––tg1282025-05-14community–source
RTX 4070 SUPERllama 7B Q4_0llama.cpp (Vulkan)112.05198.91––tg1282026-01-16community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)111.971531.85––tg1282025-05-14community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)111.97728.81––tg1282025-09-27community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)111.321420.2––tg1282025-09-11community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)111.311604.81––tg1282025-10-15community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)111.31133.3––tg1282026-08-26community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)111.133271.16––tg1282026-09-16community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)111.133271.16––tg128–community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)110.912504.25––tg1282026-03-18community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)110.81760.15––tg1282025-12-14community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)110.732372.49––tg1282025-12-14community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)110.73053.77––tg1282025-09-17community–source
RTX 4070 Tillama 7B Q4_0llama.cpp (Vulkan)110.534981.44221–tg1282026-01-17community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)110.52251.37––tg1282025-12-18community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)109.991135.88––tg1282026-08-26community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)109.923226.16––tg1282025-11-20communitycontradictionsource
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)109.923226.16––tg1282025-11-20communitycontradictionsource
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)109.814661.14––tg1282026-03-25community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)109.721382.27––tg1282026-08-26community–source
V100llama 7B Q4_0llama.cpp (CUDA)109.49380.4––tg1282026-05-26community–source
V100llama 7B Q4_0llama.cpp (CUDA)109.461156.76––tg1282026-05-26community–source
M2llama 7B Q4_0llama.cpp (Vulkan)109.21792.74––tg1282025-10-18community–source
RTX 4070 SUPERllama 7B Q4_0llama.cpp (Vulkan)108.744608.2––tg1282026-01-16community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)108.45783.94––tg1282025-09-01community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)108.454766.48––tg1282026-04-01community–source
A100llama 7B Q4_0llama.cpp (Vulkan)108.394040.77––tg1282026-08-14community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)108.01901.2––tg1282025-09-09community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)108.01901.2––tg1282025-09-09community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)107.531260.54––tg1282025-02-21community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)107.324761.91––tg1282025-01-16community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)107.161934.05––tg1282025-07-27communitycontradictionsource
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)107.111324.76––tg1282026-09-16community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)107.111324.76––tg128–community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)107.023044.46––tg1282025-05-20community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)106.993009.58––tg1282026-08-26community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)106.74753.07––tg1282026-03-25community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)106.69971.17––tg1282026-08-26community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)106.634732.96––tg1282026-04-20community–source
RX 6800llama 7B Q4_0llama.cpp (Vulkan)106.491597.95––tg1282026-10-01community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)106.151761.93––tg1282025-09-09community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)106.151761.93––tg1282025-09-09community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)106.03912.47––tg1282025-09-01community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)105.912411.47––tg1282025-12-14community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)105.642830.14––tg1282026-08-26community–source
Arc Pro B70llama 7B Q4_0llama.cpp (SYCL)105.45970.74––tg1282026-08-26community–source
M2llama 7B Q4_0llama.cpp (Vulkan)105.06796.29––tg1282025-10-18community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)104.751711.33––tg1282025-12-14community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)104.421907.81––tg1282025-11-04community–source
Quadro RTX 6000llama 7B Q4_0llama.cpp (CUDA)104.383053.96––tg1282025-09-15community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)103.98883.2––tg1282025-02-21community–source
A4500llama 7B Q4_0llama.cpp (CUDA)103.03453.1––tg1282025-08-14community–source
Quadro RTX 6000llama 7B Q4_0llama.cpp (CUDA)102.772751.18––tg1282025-09-15community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)102.661986.47––tg1282025-11-04community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)102.554336.99––tg1282025-11-15community–source
A100llama 7B Q4_0llama.cpp (Vulkan)102.463222.01––tg1282026-08-14community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)102.445009.54––tg1282026-03-06community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (ROCm)102.382982.56––tg1282026-03-18community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)102.311741.01––tg1282025-11-04community–source
M3 Ultrallama 7B Q4_0llama.cpp (Metal)102.281574.09––tg1282025-04-27community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (Vulkan)101.731465.45––tg1282025-09-11community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)101.615077.32––tg1282025-12-22community–source
RX 6800llama 7B Q4_0llama.cpp (Vulkan)101.471593.97––tg1282026-10-01community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)101.421257.98––tg1282025-01-18community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)101.351678.98––tg1282025-11-04community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)101.275055.19––tg1282025-11-26community–source
RTX 2080 Tillama 7B Q4_0llama.cpp (Vulkan)100.991936.25––tg1282025-08-27community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)100.942151.81––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)100.942151.81––tg1282025-08-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)100.621668.19––tg1282025-01-08communitycontradictionsource
GeForce RTX 5060llama 7B Q4_0llama.cpp (CUDA)100.613799.47––tg1282026-06-19community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)100.51704.79––tg1282025-09-01community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)100.321752.92––tg1282025-09-01community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)99.961601.25––tg1282026-03-01community–source
V100llama 7B Q4_0llama.cpp (CUDA)99.95977.33––tg1282026-05-26community–source
V100llama 7B Q4_0llama.cpp (CUDA)99.84354.07––tg1282026-05-26community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)99.552109.38––tg1282025-08-08community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)99.392361.1––tg1282025-08-17community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)98.891121.36––tg1282025-02-05community–source
V620llama 7B Q4_0llama.cpp (ROCm)98.782173.13––tg1282026-09-10community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)98.715025.17––tg1282026-03-01community–source
A4500llama 7B Q4_0llama.cpp (CUDA)98.452960.88––tg1282025-08-14community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Metal)98.27326.29––tg1282025-08-11community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)98.152091.34––tg1282025-05-14communitycontradictionsource
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)98.131487.1––tg1282025-05-14communitycontradictionsource
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)98.134244.92––tg1282025-11-15community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)98.11949.45––tg1282026-02-12community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)98.113840.62––tg1282026-03-06community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)97.984773.07––tg1282025-11-26community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)97.961566.38––tg1282025-01-08communitycontradictionsource
RTX 2080 Tillama 7B Q4_0llama.cpp (Vulkan)97.581888.24––tg1282025-08-27community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (ROCm)97.531598.79––tg1282025-10-15community–source
A4500llama 7B Q4_0llama.cpp (CUDA)97.322827.2––tg1282025-08-14community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)97.324613.86––tg1282026-03-06community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)97.3933.71––tg1282026-02-12community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)97.284903.51––tg1282025-11-26community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)97.275025.74––tg1282026-03-05community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)97.214944.22––tg1282025-12-22community–source
Radeon RX 9070llama 7B Q4_0llama.cpp (ROCm)97.11147.66––tg1282025-08-17community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)97.072026.84––tg1282025-08-08community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)97.013912.65––tg1282026-02-03community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)96.892145.6––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)96.892145.6––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)96.82296.72––tg1282025-08-08community–source
GeForce RTX 5060llama 7B Q4_0llama.cpp (CUDA)96.73269.08––tg1282026-06-19community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)96.651749.46––tg1282026-02-13community–source
M4 Maxllama 7B Q4_0llama.cpp (Metal)96.62823.99––tg1282025-07-15community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)96.612055.58––tg1282025-08-08community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (ROCm)96.532708.46––tg1282026-03-18community–source
GeForce RTX 3060 Tillama 7B Q4_0llama.cpp (CUDA)96.52990.99––tg1282026-09-19community–source
Radeon RX 7900 GREllama 7B Q4_0llama.cpp (ROCm)96.071456.98––tg1282025-10-15community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)96.032063.66––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)96.032063.66––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)95.992304.63––tg1282025-08-01community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (ROCm)95.992304.63––tg1282025-08-01community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)95.983848.92––tg1282026-02-03community–source
A4500llama 7B Q4_0llama.cpp (CUDA)95.873382.11––tg1282025-08-14community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)95.843731.36––tg1282026-05-08community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)95.71018.94––tg1282025-01-18community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)95.694427.42––tg1282026-05-08community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)95.611698.69––tg1282026-02-13community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)95.561533.6––tg1282025-04-12community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)95.2949.34––tg1282025-05-14communitycontradictionsource
M4 Maxllama 7B Q4_0llama.cpp (Metal)95.19949.34––tg1282025-07-15community–source
Radeon RX 7800 XTllama 7B Q4_0llama.cpp (Vulkan)95.171955.91––tg1282025-05-14communitycontradictionsource
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)95.14430.01––tg1282026-05-08community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)95.044932.17––tg1282026-03-01community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)94.81625.53––tg1282025-05-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)94.323845.97––tg1282026-03-05community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)94.34034.65––tg1282025-04-24community–source
M2 Ultra (76 GPU)llama 7B Q4_0llama.cpp (Metal)94.271238.48–––2023-11-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)93.93912.3––tg1282026-03-06community–source
NVIDIA RTX 4500llama 7B Q4_0llama.cpp (CUDA)93.886490.28––tg1282026-01-22community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)93.844443.54––tg1282025-11-26community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)93.763713.27––tg1282025-11-04communitycontradictionsource
RX 6800llama 7B Q4_0llama.cpp (ROCm)93.61738.62––tg1282026-10-01community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)93.513460.92––tg1282026-02-03community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)93.464195.53––tg1282025-08-02community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)93.423275.48––tg1282026-08-08community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)93.34163.39––tg1282025-07-30community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)93.264055.36––tg1282025-10-05community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)92.961869.09––tg1282025-05-12community–source
Radeon RX 6950 XTllama 7B Q4_0llama.cpp (ROCm)92.961869.09––tg1282025-05-12community–source
NVIDIA RTX 4500llama 7B Q4_0llama.cpp (CUDA)92.945512.71––tg1282026-01-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)92.753541.12––tg1282026-05-08community–source
RTX PRO 4000 Blackwellllama 7B Q4_0llama.cpp (CUDA)92.544078.52––tg1282026-05-13community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)92.523733.17––tg1282026-05-08community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)92.453211.59––tg1282025-12-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)92.281520.24––tg1282026-03-01community–source
GeForce RTX 3060 Tillama 7B Q4_0llama.cpp (CUDA)92.232675.22––tg1282026-09-19community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)92.193342.62––tg1282026-02-03community–source
M3 Ultra (80 GPU)llama 7B Q4_0llama.cpp (Metal)92.141471.24–––2023-11-22community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)91.86589.16––tg1282025-08-07community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)91.853720.85––tg1282025-11-04communitycontradictionsource
V620llama 7B Q4_0llama.cpp (ROCm)91.701846.20––tg1282026-09-10community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)91.59863.03––tg1282025-01-08community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)91.564073.37––tg1282025-07-30community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)91.481484.04––tg1282025-01-08communitycontradictionsource
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)91.413240.12––tg1282025-11-04communitycontradictionsource
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)90.953728.09––tg1282025-07-30community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)90.943737.25––tg1282025-08-02community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)90.782970.97149.0–tg1282025-12-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)90.613359.51––tg1282026-05-08community–source
GeForce RTX 3080llama 7B Q4_0llama.cpp (CUDA)90.62875.02––tg1282026-08-08community–source
RTX 3060 Tillama 7B Q4_0llama.cpp (Vulkan)90.352644.95––tg1282026-04-30community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)90.08723.41––tg1282025-08-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)89.94065.43––tg1282025-09-23community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)89.873917.29––tg1282025-07-30community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)89.753526.38––tg1282025-07-30community–source
RTX PRO 4000 Blackwellllama 7B Q4_0llama.cpp (CUDA)89.733628.68––tg1282026-05-13community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)89.573648.71––tg1282025-07-30community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)89.514543.73––tg1282025-11-08community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)89.174027.14––tg1282025-09-23community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)88.964029.45––tg1282025-10-05community–source
M2 Ultra (60 GPU)llama 7B Q4_0llama.cpp (Metal)88.641013.81–––2023-11-22community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)88.491889.84––tg1282025-09-09community–source
M3 Ultra (60 GPU)llama 7B Q4_0llama.cpp (Metal)88.41073.09–––2023-11-22community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (ROCm)88.311706.8––tg1282025-09-01community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)88.13893.17––tg1282026-06-13community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)88.121545.39––tg1282025-01-08communitycontradictionsource
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)87.574131.22––tg1282025-09-03community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)87.56519.14––tg1282025-05-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)87.433538.39––tg1282025-09-03community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)87.053248.01––tg1282025-11-04communitycontradictionsource
Radeon RX 6800llama 7B Q4_0llama.cpp (ROCm)86.911675.31––tg1282026-06-04community–source
RTX 3060 Tillama 7B Q4_0llama.cpp (Vulkan)86.742361.43––tg1282026-04-30community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)86.541806.18––tg1282026-06-04community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)86.47510.8––tg1282025-05-07community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)86.17522.39––tg1282025-05-07community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)86.163750.46––tg1282025-11-08community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)86.12909.86––tg1282025-05-26community–source
RX 6800llama 7B Q4_0llama.cpp (ROCm)86.031510.13––tg1282026-10-01community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)85.61843.15––tg1282025-05-12community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)85.61843.15––tg1282025-05-12community–source
Radeon RX 6950 XTllama 7B Q4_0llama.cpp (ROCm)85.491956.82––tg1282025-05-12community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)85.481963.96––tg1282025-01-18community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)85.431341.03––tg1282025-08-24community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)85.421962.01––tg1282025-05-12community–source
A4000llama 7B Q4_0llama.cpp (CUDA)85.172807.83––tg1282025-08-17community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)85.122738.43––tg1282025-05-26community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)85.041948.31––tg1282025-09-09community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (ROCm)84.881199.27––tg1282025-09-01community–source
RTX 3070llama 7B Q4_0llama.cpp (CUDA)84.622993.73––tg1282025-05-29community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)84.44928.91––tg1282026-08-27community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)84.342728.17––tg1282025-05-26community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)84.12664.82––tg1282026-08-27community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (ROCm)83.921447.07––tg1282025-08-11community–source
Radeon RX 6700 XTllama 7B Q4_0llama.cpp (Vulkan)83.881051.2––tg1282025-09-16community–source
A4000llama 7B Q4_0llama.cpp (CUDA)83.772684.06––tg1282025-08-17community–source
M1 Ultra (64 GPU)llama 7B Q4_0llama.cpp (Metal)83.731030.04–––2023-11-22community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)83.671954.15––tg1282026-06-13community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)83.632908.66––tg1282025-08-24community–source
RTX 3090llama 7B Q4_0llama.cpp (ROCm)83.55997.21––tg1282025-09-01community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)83.54713.77––tg1282025-02-08community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)83.391542.23––tg1282026-06-04community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)83.263492.22––tg1282025-04-22community–source
M4 Max (40 GPU)llama 7B Q4_0llama.cpp (Metal)83.06885.68–––2023-11-22community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)83.021824.47––tg1282025-08-15community–source
A770llama 7B Q4_0llama.cpp (SYCL)82.922416.16––tg1282025-09-22community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)82.851631.3––tg1282026-06-04community–source
Radeon RX 6950 XTllama 7B Q4_0llama.cpp (ROCm)82.731666.05––tg1282025-05-12community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)82.492015.51––tg1282026-06-04community–source
Radeon RX 6750 XTllama 7B Q4_0llama.cpp (Vulkan)82.29997.05––tg1282025-06-10community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)82.26643.37––tg1282025-08-11community–source
Arc Pro B70llama 7B Q4_0llama.cpp (Vulkan)82.192461.92––tg1282026-07-17community–source
Radeon RX 6750 XTllama 7B Q4_0llama.cpp (Vulkan)81.981040.58––tg1282025-06-10community–source
Radeon RX 6700 XTllama 7B Q4_0llama.cpp (Vulkan)81.861010.9––tg1282025-09-16community–source
A770llama 7B Q4_0llama.cpp (SYCL)81.62417.17––tg1282025-09-22community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)81.483211.73––tg1282025-04-22community–source
NVIDIA RTX 3500llama 7B Q4_0llama.cpp (CUDA)81.173353.26––tg1282025-09-20community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)80.811841.07––tg1282025-05-26community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)80.592012.88––tg1282025-11-23community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)80.451250.68––tg1282025-08-15community–source
Arc Pro B70llama 7B Q4_0llama.cpp (Vulkan)80.392531.36––tg1282026-07-17community–source
A4000llama 7B Q4_0llama.cpp (CUDA)80.392835.78––tg1282026-01-08community–source
NVIDIA RTX 3500llama 7B Q4_0llama.cpp (CUDA)80.183757.32––tg1282025-09-20community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)80.031869.16––tg1282026-08-27community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)80.03570.18––tg1282026-08-27community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (ROCm)79.781382.74––tg1282026-06-04community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)79.6444.07––tg1282025-05-07communitycontradictionsource
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)79.35441.9––tg1282025-05-07communitycontradictionsource
M1 Ultrallama 7B Q8_0llama.cpp (Metal)79.131167.67––tg1282026-08-28community–source
RTX 3070llama 7B Q4_0llama.cpp (Vulkan)78.712113.02––tg1282025-05-29community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)78.3570.06––tg1282025-05-07communitycontradictionsource
Radeon RX 6750 XTllama 7B Q4_0llama.cpp (Vulkan)78.19920.41––tg1282026-07-20community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)78.192452.67––tg1282025-11-23community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)78.151737.0––tg1282026-06-04community–source
A4000llama 7B Q4_0llama.cpp (CUDA)78.122523.51––tg1282026-01-08community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)78.13504.54––tg1282025-07-30community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)77.921851.35––tg1282026-06-04community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)77.841615.12––tg1282025-05-12community–source
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)77.741970.1––tg1282026-05-25community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)77.65652.15152.0–tg1282026-01-17community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)77.0172.66––tg1282025-09-24community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)76.922407.67––tg1282025-08-21community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)76.831841.31––tg1282026-06-04community–source
A4000llama 7B Q4_0llama.cpp (Vulkan)76.072482.74––tg1282026-01-08community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)75.941815.7––tg1282025-11-23community–source
Radeon RX 6950 XTllama 7B Q4_0llama.cpp (ROCm)75.81730.22––tg1282025-05-12community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)75.751479.07––tg1282025-07-30community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)75.582407.93––tg1282026-08-21community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)75.572137.5––tg1282025-08-21community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (ROCm)75.241736.24––tg1282025-05-12community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)75.172118.17––tg1282025-11-23community–source
M4 Maxllama 7B Q4_0llama.cpp (Vulkan)75.02724.77––tg1282025-07-15community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)74.982401.11––tg1282025-08-23community–source
M1 Ultra (48 GPU)llama 7B Q4_0llama.cpp (Metal)74.93772.24–––2023-11-22community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)74.83386.08––tg1282025-07-30community–source
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)74.761840.52––tg1282026-05-25community–source
A5500llama 7B Q4_0llama.cpp (CUDA)74.682180.23––tg1282026-05-25community–source
A5500llama 7B Q4_0llama.cpp (CUDA)74.662338.14––tg1282026-05-25community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)74.661803.65––tg1282025-08-02community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)74.523060.78––tg1282025-07-30community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)74.473226.8––tg1282025-07-30community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)74.43538.31––tg1282026-01-29community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)74.241543.17––tg1282025-07-30community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)74.013228.15––tg1282026-10-02community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)73.862311.69––tg1282026-02-28community–source
Arc B580llama 7B Q4_0llama.cpp (SYCL)73.762063.52––tg1282026-05-27community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)73.752343.25––tg1282025-08-18community–source
A4000llama 7B Q4_0llama.cpp (Vulkan)73.742248.14––tg1282026-01-08community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)73.742346.54––tg1282026-02-19community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)73.711685.14––tg1282025-12-03community–source
M5 Ultrallama 7B 6.74BF16llama.cpp (Metal)73.65207.14––tg1282026-09-30community–source
RTX 3090llama 7B Q4_0llama.cpp (ROCm)73.58426.56––tg1282025-09-01community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)73.011929.93––tg1282026-06-04community–source
A770llama 7B Q4_0llama.cpp (SYCL)72.542564.25––tg1282025-09-21community–source
M5 Maxllama 7B Q8_0llama.cpp (Metal)72.523219.94––tg1282026-06-22community–source
M5 Max (40 GPU)llama 7B Q8_0llama.cpp (Metal)72.423143.81–––2023-11-22community–source
M5 Maxllama 7B Q8_0llama.cpp (Metal)72.423143.81––tg1282026-08-25community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)72.39111.82––tg1282025-10-26community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)72.31118.09––tg1282025-10-26community–source
M4llama 7B Q4_0llama.cpp (Metal)72.26762.81––tg1282025-05-20community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)72.241840.64––tg1282025-09-14community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)72.092111.39––tg1282026-08-21community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)72.0411.98––tg1282026-02-28community–source
Arc B580llama 7B Q4_0llama.cpp (Vulkan)72.0419.49––tg1282025-09-20community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)71.96116.28––tg1282025-10-26community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)71.952056.43––tg1282025-09-14community–source
Arc A770llama 7B Q4_0llama.cpp (ROCm)71.931357.35––tg1282025-07-30community–source
M4 Maxllama 7B Q4_0llama.cpp (Vulkan)71.79557.46––tg1282025-07-15community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)71.66474.89––tg1282025-09-23community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)71.661921.04––tg1282025-08-18community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)71.58492.13––tg1282025-09-23community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)71.5292.18––tg1282025-09-24community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)71.452093.56––tg1282026-02-28community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)71.421572.29––tg1282026-06-04community–source
2x V620llama 7B Q4_0llama.cpp (ROCm)71.382146.36––tg1282026-09-10community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)71.37612.47––tg1282025-04-12community–source
A770llama 7B Q4_0llama.cpp (SYCL)71.322545.16––tg1282025-09-21community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)71.172239.81––tg1282025-08-18community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)71.142020.41––tg1282025-08-23community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)71.072844.8––tg1282026-10-02community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (ROCm)70.831256.86––tg1282025-08-02community–source
RTX 3060llama 7B Q4_0llama.cpp (CUDA)70.782229.42––tg1282025-06-17community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)70.73529.69––tg1282026-01-29community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)70.54274.76––tg1282026-01-17community–source
A770llama 7B Q4_0llama.cpp (SYCL)70.542498.81––tg1282026-01-28community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)70.542141.67––tg1282025-06-13community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)70.521915.41––tg1282025-06-13community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)70.522070.83––tg1282026-02-19community–source
A770llama 7B Q4_0llama.cpp (SYCL)70.182478.2––tg1282025-04-01community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)70.183014.51––tg1282026-10-02community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)70.17312.02––tg1282025-01-08community–source
Arc B580llama 7B Q4_0llama.cpp (Vulkan)70.14620.94––tg1282025-09-20community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)70.13439.42––tg1282025-01-12community–source
M4 Max (32 GPU)llama 7B Q4_0llama.cpp (Metal)69.95713.93–––2023-11-22community–source
A770llama 7B Q4_0llama.cpp (SYCL)69.732512.42––tg1282026-01-28communityoutliersource
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)69.461626.71––tg1282026-06-04community–source
Arc A770llama 7B Q4_0llama.cpp (ROCm)69.27987.41––tg1282025-07-30community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)69.14605.4––tg1282025-01-15community–source
GeForce RTX 3060llama 7B Q4_0llama.cpp (CUDA)69.012028.81––tg1282025-08-18community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)68.674727.48––tg1282025-11-26communitycontradictionsource
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)68.55522.36126–tg1282026-01-17community–source
RTX 3060llama 7B Q4_0llama.cpp (CUDA)68.382020.4––tg1282025-06-17community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)68.341122.04––tg1282025-12-03community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)68.071621.0––tg1282026-06-04communitycontradictionsource
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)67.982251.85––tg1282026-02-15community–source
Quadro RTX 4000llama 7B Q4_0llama.cpp (CUDA)67.621662.8––tg1282025-11-21community–source
M5 Prollama 7B Q4_0llama.cpp (Metal)67.591340.44––tg1282026-09-24community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)67.581419.67––tg1282025-09-14community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (ROCm)67.55354.17––tg1282025-01-12community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)67.53609.57––tg1282025-01-15community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)67.272641.07––tg1282026-10-02community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)67.111826.49––tg1282025-06-17community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)66.81659.07––tg1282025-05-07community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)66.67446.42––tg1282026-02-28community–source
M2 Ultra (76 GPU)llama 7B Q8_0llama.cpp (Metal)66.641248.59–––2023-11-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)66.59572.26––tg1282026-05-08communitycontradictionsource
M2llama 7B Q4_0llama.cpp (Vulkan)66.48685.51––tg1282025-11-26community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)66.482534.4––tg1282025-12-15community–source
M5 Pro (20 GPU)llama 7B Q4_0llama.cpp (Metal)66.331620.64–––2023-11-22community–source
M5 Prollama 7B Q4_0llama.cpp (Metal)66.331620.64––tg1282026-08-31community–source
M2llama 7B Q4_0llama.cpp (Vulkan)66.32683.61––tg1282025-11-28community–source
M3 Max (40 GPU)llama 7B Q4_0llama.cpp (Metal)66.31759.7–––2023-11-22community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)66.273585.2––tg1282025-11-26communitycontradictionsource
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)66.121715.17––tg1282025-08-23community–source
M2 Max (38 GPU)llama 7B Q4_0llama.cpp (Metal)65.95671.31–––2023-11-22community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)65.941667.43––tg1282026-06-04community–source
A100llama 7B Q4_0llama.cpp (CUDA)65.91653.01––tg1282026-05-13communitycontradictionsource
Arc B580llama 7B Q4_0llama.cpp (SYCL)65.671721.91––tg1282026-05-27communitycontradictionsource
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)65.631730.33––tg1282025-08-23community–source
Quadro RTX 4000llama 7B Q4_0llama.cpp (CUDA)65.621536.89––tg1282025-11-21community–source
A100llama 7B Q4_0llama.cpp (CUDA)65.48626.32––tg1282026-05-13communitycontradictionsource
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)65.421479.27––tg1282025-09-14community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)65.362037.24––tg1282026-02-15community–source
2x V620llama 7B Q4_0llama.cpp (ROCm)64.851801.75––tg1282026-09-10community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)64.761681.81––tg1282025-06-17community–source
RTX 2070 Superllama 7B Q4_0llama.cpp (Vulkan)64.641199.13––tg1282025-02-07community–source
Radeon RX 6650 XTllama 7B Q4_0llama.cpp (Vulkan)64.531088.9––tg1282025-11-25community–source
GeForce RTX 4060 Tillama 7B Q4_0llama.cpp (CUDA)64.033803.45––tg1282025-08-02community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (ROCm)63.982427.16––tg1282025-12-15community–source
M3 Ultra (80 GPU)llama 7B Q8_0llama.cpp (Metal)63.931487.51–––2023-11-22community–source
M5 Prollama 7B Q4_0llama.cpp (Metal)63.861350.43––tg1282026-04-14community–source
GeForce RTX 4060 Tillama 7B Q4_0llama.cpp (CUDA)63.863394.63––tg1282025-08-02community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)63.79518.53––tg1282026-05-08communitycontradictionsource
M5 Prollama 7B Q4_0llama.cpp (Metal)63.741135.71––tg1282026-04-14community–source
RTX 3070llama 7B Q4_0llama.cpp (Vulkan)63.641689.4––tg1282025-10-20community–source
M3 Ultra (60 GPU)llama 7B Q8_0llama.cpp (Metal)63.551085.76–––2023-11-22community–source
M2llama 7B Q4_0llama.cpp (Vulkan)63.16678.14––tg1282025-11-26community–source
M2llama 7B Q4_0llama.cpp (Vulkan)63.1675.02––tg1282025-11-28community–source
RTX 3070llama 7B Q4_0llama.cpp (Vulkan)62.921832.07––tg1282025-10-20community–source
M2llama 7B Q4_0llama.cpp (Vulkan)62.7553.16––tg1282025-11-26community–source
GeForce RTX 3070llama 7B Q4_0llama.cpp (CUDA)62.171735.28––tg1282025-10-17community–source
RTX 3080llama 7B Q4_0llama.cpp (Vulkan)62.161706.07––tg1282024-12-18community–source
M2 Ultra (60 GPU)llama 7B Q8_0llama.cpp (Metal)62.141003.16–––2023-11-22community–source
Radeon RX 6650 XTllama 7B Q4_0llama.cpp (Vulkan)62.141029.52––tg1282025-11-25community–source
V100llama 7B Q4_0llama.cpp (CUDA)62.06788.45––tg1282025-11-26community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)62.052199.15––tg1282026-02-03community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (ROCm)62.02314.17––tg1282025-01-12community–source
P100llama 7B Q4_0llama.cpp (CUDA)61.99787.36––tg1282025-11-26community–source
NVIDIA RTX 4000llama 7B Q4_0llama.cpp (CUDA)61.832779.77––tg1282025-09-30community–source
NVIDIA RTX 4000llama 7B Q4_0llama.cpp (CUDA)61.373171.86––tg1282025-09-30community–source
M1 Max (32 GPU)llama 7B Q4_0llama.cpp (Metal)61.19530.06–––2023-11-22community–source
Radeon RX 6750 XTllama 7B Q4_0llama.cpp (Vulkan)61.08525.04––tg1282026-07-20community–source
M2 Max (30 GPU)llama 7B Q4_0llama.cpp (Metal)60.99537.6–––2023-11-22community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (Vulkan)60.881353.47––tg1282026-07-30community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)60.393321.91––tg1282026-07-14community–source
GB10llama 7B Q4_0llama.cpp (CUDA)60.243310.48––tg1282026-01-20community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)60.18721.44––tg1282025-09-24community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)60.02689.59––tg1282026-02-03community–source
RTX 4060llama 7B Q4_0llama.cpp (Vulkan)60.012358.03––tg1282025-11-04community–source
M1 Ultra (64 GPU)llama 7B Q8_0llama.cpp (Metal)59.871042.95–––2023-11-22community–source
GeForce RTX 3070llama 7B Q4_0llama.cpp (CUDA)59.721729.14––tg1282025-10-17community–source
GB10llama 7B Q4_0llama.cpp (CUDA)59.624016.46––tg1282026-03-30community–source
GeForce RTX 4060llama 7B Q4_0llama.cpp (CUDA)59.552676.6––tg1282026-02-11community–source
RTX 4060llama 7B Q4_0llama.cpp (Vulkan)59.532135.66––tg1282025-11-04community–source
GeForce RTX 3070llama 7B Q4_0llama.cpp (CUDA)59.521918.79––tg1282025-10-17community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)59.36488.06––tg1282025-09-01community–source
Radeon RX 6650 XTllama 7B Q4_0llama.cpp (Vulkan)59.22735.64––tg1282025-06-10community–source
Radeon RX 6800 XTllama 7B Q4_0llama.cpp (Vulkan)59.22484.37––tg1282025-09-01community–source
Radeon RX 6650 XTllama 7B Q4_0llama.cpp (Vulkan)59.18730.64––tg1282025-06-10community–source
GeForce RTX 4060llama 7B Q4_0llama.cpp (CUDA)59.152928.74––tg1282026-02-11community–source
GB10llama 7B Q4_0llama.cpp (CUDA)59.063306.61––tg1282026-03-30community–source
GB10llama 7B Q4_0llama.cpp (CUDA)59.053954.97––tg1282026-03-30community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)59.02586.16––tg1282025-05-10community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (ROCm)58.731308.31––tg1282026-07-14community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)58.512789.51––tg1282026-07-14community–source
M2llama 7B Q4_0llama.cpp (Vulkan)58.4753.1––tg1282025-11-26community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)58.44632.88––tg1282025-05-10community–source
GB10llama 7B Q4_0llama.cpp (CUDA)58.383261.99––tg1282026-03-30community–source
P100llama 7B Q4_0llama.cpp (CUDA)58.35760.8––tg1282025-11-26community–source
V100llama 7B Q4_0llama.cpp (CUDA)58.33767.12––tg1282025-11-26community–source
GeForce RTX 4060llama 7B Q4_0llama.cpp (CUDA)58.222466.31––tg1282026-02-11community–source
GB10llama 7B Q4_0llama.cpp (CUDA)58.143861.19––tg1282026-02-16community–source
GB10llama 7B Q4_0llama.cpp (CUDA)58.133241.51––tg1282026-02-16community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (Vulkan)58.031167.2––tg1282026-07-30community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (ROCm)58.01312.19––tg1282026-07-30community–source
GeForce RTX 4060llama 7B Q4_0llama.cpp (CUDA)57.872464.58––tg1282026-02-11community–source
GeForce RTX 3070llama 7B Q4_0llama.cpp (CUDA)57.361904.62––tg1282025-10-17community–source
GB10llama 7B Q4_0llama.cpp (CUDA)57.213062.31––tg1282025-10-15community–source
GB10llama 7B Q4_0llama.cpp (CUDA)56.743661.37––tg1282025-10-15community–source
M3 Max (30 GPU)llama 7B Q4_0llama.cpp (Metal)56.58567.59–––2023-11-22community–source
Radeon RX 7900 XTXllama 7B Q4_0llama.cpp (Vulkan)56.351727.1––tg1282026-08-23community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)56.111024.38––tg1282026-02-15community–source
M1 Ultra (48 GPU)llama 7B Q8_0llama.cpp (Metal)55.69783.45–––2023-11-22community–source
Radeon RX 7900 XTllama 7B Q4_0llama.cpp (Vulkan)55.41722.38––tg1282025-09-24community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)55.321147.07––tg1282025-12-11community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)55.21028.16––tg1282026-02-15community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)54.961651.6––tg1282025-12-30community–source
GB10llama 7B Q4_0llama.cpp (CUDA)54.913920.63––tg1282026-03-30community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)54.84418.46––tg1282025-12-30community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)54.761135.17––tg1282025-11-19community–source
M1 Max (24 GPU)llama 7B Q4_0llama.cpp (Metal)54.61400.26–––2023-11-22community–source
GB10llama 7B Q4_0llama.cpp (CUDA)54.393293.78––tg1282026-03-30community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)54.38979.76––tg1282026-09-17community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)54.381704.3––tg1282026-10-02community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)54.281298.03––tg1282025-01-21community–source
M4 Max (40 GPU)llama 7B Q8_0llama.cpp (Metal)54.05891.94–––2023-11-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)53.961037.02––tg1282026-05-08communitycontradictionsource
Radeon RX 6600 XTllama 7B Q4_0llama.cpp (Vulkan)53.92574.65––tg1282025-01-14community–source
RTX 4060llama 7B Q4_0llama.cpp (Vulkan)53.81908.24––tg1282026-10-10community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)53.7612.33––tg1282025-08-01community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)53.65428.35––tg1282026-05-08communitycontradictionsource
GB10llama 7B Q4_0llama.cpp (Vulkan)53.643279.89––tg1282026-03-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)53.591288.96––tg1282025-09-21community–source
Arc Pro B60llama 7B Q4_0llama.cpp (Vulkan)53.561581.29––tg1282026-06-04communitycontradictionsource
Radeon RX 7600llama 7B Q4_0llama.cpp (ROCm)53.481076.93––tg1282026-07-30community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)53.371083.51––tg1282025-11-12community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (ROCm)53.271006.77––tg1282026-07-14community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)53.24808.76––tg1282026-03-21community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)53.213219.57––tg1282026-03-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)53.211120.83––tg1282025-10-24community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)53.17804.64––tg1282026-07-18community–source
A770llama 7B Q4_0llama.cpp (Vulkan)53.071119.68––tg1282026-03-24community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)53.02840.85––tg1282026-02-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)53.01357.07––tg1282025-09-21community–source
A770llama 7B Q4_0llama.cpp (SYCL)52.982173.52––tg1282025-07-30community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)52.85416.06––tg1282026-05-08communitycontradictionsource
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)52.84606.24––tg1282025-08-01community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)52.8544.83––tg1282026-02-15community–source
A770llama 7B Q4_0llama.cpp (SYCL)52.76916.88––tg1282026-01-28communitycontradictionsource
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)52.691039.52––tg1282026-02-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.691243.31––tg1282025-10-24community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)52.66428.54––tg1282025-11-26community–source
A770llama 7B Q4_0llama.cpp (SYCL)52.621866.14––tg1282025-09-21communitycontradictionsource
RTX 4060llama 7B Q4_0llama.cpp (Vulkan)52.581793.16––tg1282026-10-10community–source
A770llama 7B Q4_0llama.cpp (Vulkan)52.561073.85––tg1282026-03-24community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.51264.6––tg1282025-10-24community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)52.48296.62––tg1282025-11-04community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)52.47410.19––tg1282025-12-21community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)52.282737.79––tg1282026-03-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.28972.06––tg1282025-10-23community–source
A770llama 7B Q4_0llama.cpp (SYCL)52.262185.36––tg1282025-07-30community–source
P100llama 7B Q4_0llama.cpp (CUDA)52.241076.52––tg1282025-11-27community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.231044.51––tg1282025-12-11community–source
A770llama 7B Q4_0llama.cpp (Vulkan)52.04986.03––tg1282025-09-21community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.031305.74––tg1282025-10-24community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)52.01869.4––tg1282025-09-10community–source
A2000llama 7B Q4_0llama.cpp (CUDA)51.961411.42––tg1282025-08-23community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.89803.41––tg1282025-08-17community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.811487.08––tg1282026-10-02community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.71951.25––tg1282025-09-10community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)51.72710.58––tg1282026-03-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.681041.42––tg1282025-11-19community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.671023.92––tg1282025-11-12community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.621054.53––tg1282025-10-23community–source
P100llama 7B Q4_0llama.cpp (CUDA)51.541006.05––tg1282025-11-27community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)51.431185.04––tg1282025-11-04community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)51.42494.8––tg1282025-05-06community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)51.36442.71––tg1282025-07-22community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)51.36251.11––tg1282025-11-04community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)51.331456.15––tg1282026-10-02community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (CUDA)51.3913.16––tg1282026-05-08communitycontradictionsource
A2000llama 7B Q4_0llama.cpp (CUDA)51.211259.66––tg1282025-08-23community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)51.14927.96––tg1282025-08-17community–source
P100llama 7B Q4_0llama.cpp (CUDA)51.08735.19––tg1282025-08-23community–source
A770llama 7B Q4_0llama.cpp (Vulkan)50.9903.49––tg1282025-09-21community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)50.82665.61––tg1282026-01-20community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.751117.55––tg1282025-11-15community–source
M4 Pro (20 GPU)llama 7B Q4_0llama.cpp (Metal)50.74439.78–––2023-11-22community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.691118.89––tg1282025-11-15community–source
P100llama 7B Q4_0llama.cpp (CUDA)50.66492.2––tg1282025-11-27community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)50.63761.89––tg1282026-03-21community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)50.63483.22––tg1282025-05-21community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.591012.79––tg1282025-11-26community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)50.57758.5––tg1282026-07-18community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.571576.59––tg1282025-12-11community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)50.49288.51––tg1282025-09-21community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)50.47926.02––tg1282025-11-04community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.41488.56––tg1282026-02-25community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)50.39244.24––tg1282025-09-01community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.3998.73––tg1282025-12-20community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)50.291464.06––tg1282025-12-15community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)50.23439.87––tg1282025-11-26community–source
P100llama 7B Q4_0llama.cpp (CUDA)50.2703.27––tg1282025-08-23community–source
P100llama 7B Q4_0llama.cpp (CUDA)50.17476.77––tg1282025-11-27community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)50.11404.04––tg1282025-12-21community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)49.76331.58––tg1282025-03-10community–source
A770llama 7B Q4_0llama.cpp (Vulkan)49.66372.22––tg1282025-09-21community–source
M4 Pro (16 GPU)llama 7B Q4_0llama.cpp (Metal)49.64364.06–––2023-11-22community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)49.64913.95––tg1282025-09-21community–source
Radeon RX 7600Sllama 7B Q4_0llama.cpp (ROCm)49.62915.55––tg1282026-05-09community–source
P102-100llama 7B Q4_0llama.cpp (CUDA)49.591032.13––tg1282026-10-01community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)49.58663.96––tg1282025-07-22community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)49.55762.02––tg1282025-09-01community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)49.51223.09––tg1282026-04-03community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)49.442021.05––tg1282026-02-03community–source
A770llama 7B Q4_0llama.cpp (Vulkan)49.43725.31––tg1282025-05-26community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)49.34295.72––tg1282025-09-21community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)49.11143.51––tg1282026-08-11community–source
P102-100llama 7B Q4_0llama.cpp (CUDA)49.081029.43––tg1282026-10-01community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)49.06784.16––tg1282025-07-17community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)49.06784.16––tg1282025-07-22community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)49.031059.76––tg1282025-03-20community–source
A770llama 7B Q4_0llama.cpp (Vulkan)49.02603.46––tg1282025-04-03community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)48.97366.08––tg1282025-09-23community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)48.71861.99––tg1282025-07-17community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)48.71861.99––tg1282025-07-22community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)48.64325.3––tg1282025-11-26community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (ROCm)48.581099.64––tg1282025-08-01community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)48.52617.85––tg1282025-08-15community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)48.46904.11––tg1282025-09-21community–source
A770llama 7B Q4_0llama.cpp (Vulkan)48.451116.8––tg1282025-04-01community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)48.422692.06––tg1282026-03-15community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)48.31420.83––tg1282025-05-21community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)48.31622.72––tg1282025-08-15community–source
A770llama 7B Q4_0llama.cpp (Vulkan)48.3363.82––tg1282025-09-21community–source
Radeon RX 6600Mllama 7B Q4_0llama.cpp (Vulkan)48.21605.59––tg1282025-04-10community–source
A770llama 7B Q4_0llama.cpp (Vulkan)48.17327.58––tg1282025-05-26community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)47.97351.36––tg1282025-09-23community–source
Radeon RX 6600 XTllama 7B Q4_0llama.cpp (Vulkan)47.94428.92––tg1282025-02-08community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)47.94717.53––tg1282025-05-21community–source
M1 Ultrallama 7B F16llama.cpp (Metal)47.821294.31––tg1282026-08-28community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)47.81826.15––tg1282026-09-17community–source
Arc Pro B50llama 7B Q4_0llama.cpp (SYCL)47.8397.73––tg1282026-05-28community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)47.691122.53––tg1282026-05-30community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (ROCm)47.651199.16––tg1282025-08-01community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)47.552731.93––tg1282026-03-14community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)47.49615.03––tg1282025-06-27community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)47.47380.87––tg1282025-03-10community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)47.361313.76––tg1282026-10-02community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)47.35511.59––tg1282026-02-03community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)46.98694.27––tg1282025-05-21community–source
P102-100llama 7B Q4_0llama.cpp (CUDA)46.77935.02––tg1282026-10-01community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)46.691121.72––tg1282025-08-17communitycontradictionsource
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.681009.02––tg1282025-11-15community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)46.673137.97––tg1282026-03-15community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.651004.84––tg1282025-11-15community–source
GB10llama 7B Q4_0llama.cpp (Vulkan)46.613171.16––tg1282026-03-14community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.51908.35––tg1282025-11-26community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.461352.52––tg1282025-12-11community–source
P102-100llama 7B Q4_0llama.cpp (CUDA)46.45927.54––tg1282026-10-01community–source
T4llama 7B Q4_0llama.cpp (CUDA)46.381219.06––tg1282025-08-14community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)46.31186.45––tg1282026-05-30community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)46.18311.9––tg1282025-11-26community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.15843.04––tg10242026-04-03community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (Vulkan)46.091347.94––tg1282025-08-17community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)46.011201.51––tg1282026-02-25community–source
Arc Pro B50llama 7B Q4_0llama.cpp (SYCL)45.92424.5––tg1282026-05-28community–source
Arc Pro B50llama 7B Q4_0llama.cpp (SYCL)45.74397.09––tg1282026-05-28community–source
A2000llama 7B Q4_0llama.cpp (CUDA)45.691361.85––tg1282025-08-23community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)45.531168.63––tg1282025-12-15community–source
A2000llama 7B Q4_0llama.cpp (CUDA)45.521245.19––tg1282025-08-23community–source
A770llama 7B Q4_0llama.cpp (Vulkan)45.22314.24––tg1282025-01-10community–source
Radeon RX 7600Sllama 7B Q4_0llama.cpp (ROCm)45.02783.23––tg1282026-05-09community–source
M4llama 7B Q8_0llama.cpp (Metal)44.89702.39––tg1282025-05-20community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)44.4995.52––tg1282025-01-08community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)44.151106.09––tg1282026-04-03community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)44.1658.95––tg1282025-06-13community–source
T4llama 7B Q4_0llama.cpp (CUDA)44.031309.73––tg1282025-08-14community–source
Arc A750llama 7B Q4_0llama.cpp (Vulkan)43.96303.37––tg1282025-11-07community–source
Arc Pro B60llama 7B Q4_0llama.cpp (SYCL)43.93399.3––tg1282026-01-17community–source
M4 Max (32 GPU)llama 7B Q8_0llama.cpp (Metal)43.87718.56–––2023-11-22community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)43.792784.46––tg1282025-07-30community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (ROCm)43.791063.6––tg1282026-08-11community–source
RTX 4050llama 7B Q4_0llama.cpp (CUDA)43.721725.85––tg1282025-01-13community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)43.38285.36––tg1282025-08-17community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)43.35211.25––tg1282026-07-09community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)43.33330.84––tg1282025-11-26community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)43.18822.07––tg1282026-08-31community–source
M3 Max (40 GPU)llama 7B Q8_0llama.cpp (Metal)42.75757.64–––2023-11-22community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)42.7469.46––tg1282025-08-17community–source
Arc A750llama 7B Q4_0llama.cpp (Vulkan)42.661075.94––tg1282025-11-07community–source
RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)42.472718.32––tg1282025-07-30community–source
M3 Ultra (60 GPU)llama 7B F16llama.cpp (Metal)42.241121.8–––2023-11-22community–source
Arc A750llama 7B Q4_0llama.cpp (Vulkan)42.15312.51––tg1282025-08-18community–source
RTX 4050llama 7B Q4_0llama.cpp (Vulkan)41.891154.28––tg1282025-01-13community–source
M2 Max (38 GPU)llama 7B Q8_0llama.cpp (Metal)41.83677.91–––2023-11-22community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)41.82581.78––tg1282026-05-30community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)41.82581.78––tg1282026-05-28community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)41.48321.57––tg1282025-11-26community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)41.48697.13––tg1282026-08-31community–source
Arc A750llama 7B Q4_0llama.cpp (Vulkan)41.43665.38––tg1282025-08-18community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)41.19257.53––tg1282025-07-30community–source
M2 Ultra (76 GPU)llama 7B F16llama.cpp (Metal)41.021401.85–––2023-11-22community–source
Arc B50llama 7B Q4_0llama.cpp (Vulkan)41.02132.48––tg1282025-12-06community–source
A750llama 7B Q4_0llama.cpp (OpenCL)40.7234.63––tg1282025-11-07community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)40.621243.46––tg1282026-05-31community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)40.56737.29––tg1282025-07-30community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)40.4106.39––tg1282025-10-27community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)40.37217.77––tg1282025-11-26community–source
M1 Max (32 GPU)llama 7B Q8_0llama.cpp (Metal)40.2537.37–––2023-11-22community–source
Radeon RX 570 Sllama 7B Q4_0llama.cpp (Vulkan)40.18229.23––tg1282026-04-25community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)40.13590.01––tg1282026-05-30community–source
Arc Pro B50llama 7B Q4_0llama.cpp (Vulkan)40.13590.01––tg1282026-05-28community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)40.11207.16––tg1282026-07-09community–source
Arc B50llama 7B Q4_0llama.cpp (Vulkan)39.99193.5––tg1282025-12-06community–source
M2 Max (30 GPU)llama 7B Q8_0llama.cpp (Metal)39.97540.15–––2023-11-22community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)39.93107.79––tg1282025-10-27community–source
M2 Ultra (60 GPU)llama 7B F16llama.cpp (Metal)39.861128.59–––2023-11-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)39.85206.61––tg1282026-05-27community–source
M3 Ultra (80 GPU)llama 7B F16llama.cpp (Metal)39.781538.34–––2023-11-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)39.75207.79––tg1282026-05-27community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)39.41661.92––tg1282025-04-12community–source
Radeon RX 580 Sllama 7B Q4_0llama.cpp (Vulkan)39.32258.03––tg1282025-05-12community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)39.131209.43––tg1282026-05-31community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)39.09692.82––tg1282026-05-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)38.96198.66––tg1282026-05-27community–source
M5 Prollama 7B Q8_0llama.cpp (Metal)38.941302.59––tg1282026-09-24community–source
M5 Pro (20 GPU)llama 7B Q8_0llama.cpp (Metal)38.921553.86–––2023-11-22community–source
M5 Prollama 7B Q8_0llama.cpp (Metal)38.921553.86––tg1282026-08-31community–source
M2 Pro (19 GPU)llama 7B Q4_0llama.cpp (Metal)38.86341.19–––2023-11-22community–source
Arc A770llama 7B Q4_0llama.cpp (Vulkan)38.8255.59––tg1282025-07-30communitycontradictionsource
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)38.69198.3––tg1282026-05-27community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)38.63218.07––tg1282025-08-23community–source
GeForce RTX 3050llama 7B Q4_0llama.cpp (CUDA)38.511258.25––tg1282026-07-15community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)38.371649.97––tg1282025-12-30community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)38.31317.42––tg1282025-12-30community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)38.21192.0––tg1282025-09-08community–source
Radeon RX 580 Sllama 7B Q4_0llama.cpp (Vulkan)38.2222.98––tg1282025-06-02community–source
A770llama 7B Q4_0llama.cpp (SYCL)38.1845.83––tg1282026-01-28communitycontradictionsource
Arc A770llama 7B Q4_0llama.cpp (Vulkan)38.07706.67––tg1282025-07-30communitycontradictionsource
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)37.94401.03––tg1282026-02-15communitycontradictionsource
M2 Pro (16 GPU)llama 7B Q4_0llama.cpp (Metal)37.87294.24–––2023-11-22community–source
M1 Max (24 GPU)llama 7B Q8_0llama.cpp (Metal)37.81405.87–––2023-11-22community–source
M5 Prollama 7B Q8_0llama.cpp (Metal)37.781304.66––tg1282026-04-14community–source
M5 Prollama 7B Q8_0llama.cpp (Metal)37.7964.18––tg1282026-04-14community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)37.5923.61––tg1282025-10-26communitycontradictionsource
Radeon RX 570 Sllama 7B Q4_0llama.cpp (Vulkan)37.44226.11––tg1282026-04-25community–source
A750llama 7B Q4_0llama.cpp (OpenCL)37.42162.56––tg1282025-11-07community–source
GeForce RTX 3050llama 7B Q4_0llama.cpp (CUDA)37.391147.01––tg1282026-07-15community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)37.3321.66––tg1282025-10-26communitycontradictionsource
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)37.29214.61––tg1282025-11-26communitycontradictionsource
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)37.28197.26––tg1282026-02-25community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)37.23194.52––tg1282025-10-24community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)37.15182.13––tg1282025-09-08community–source
M5 Max (40 GPU)llama 7B F16llama.cpp (Metal)37.113158.49–––2023-11-22community–source
M5 Maxllama 7B F16llama.cpp (Metal)37.113158.49––tg1282026-08-25community–source
M1 Ultra (64 GPU)llama 7B F16llama.cpp (Metal)37.011168.89–––2023-11-22community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)36.82315.39––tg1282025-06-18community–source
Radeon RX 6900 XTllama 7B Q4_0llama.cpp (Vulkan)36.7223.9––tg1282025-10-26communitycontradictionsource
Arc A750llama 7B Q4_0llama.cpp (SYCL)36.641616.11––tg1282025-01-08community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)36.49201.82––tg1282025-10-24community–source
M1 Pro (16 GPU)llama 7B Q4_0llama.cpp (Metal)36.41266.25–––2023-11-22community–source
Arc A750llama 7B Q4_0llama.cpp (SYCL)36.21631.21––tg1282025-11-07community–source
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)36.11549.43––tg1282026-05-25community–source
M2 Ultrallama 7B Q4_0llama.cpp (Vulkan)36.07198.83––tg1282025-11-25community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)35.79180.49––tg1282025-11-26community–source
M6llama 7B 6.74BQ4_0llama.cpp (Metal)35.74988.93––tg1282026-10-01community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)35.59157.61––tg1282025-11-26community–source
M1 Pro (14 GPU)llama 7B Q4_0llama.cpp (Metal)35.52232.55–––2023-11-22community–source
M5 Maxllama 7B F16llama.cpp (Metal)35.393189.9––tg1282026-06-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)35.14197.94––tg1282025-08-23community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)34.89185.73––tg1282025-08-22community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)34.72389.52––tg1282025-06-13community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)34.71248.66––tg1282025-09-21community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)34.71189.5––tg1282026-05-08communitycontradictionsource
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)34.71545.12––tg1282026-05-25community–source
T4llama 7B Q4_0llama.cpp (Vulkan)34.66734.28––tg1282025-07-12community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)34.65179.01––tg1282025-08-22community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)34.47912.36––tg1282026-05-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)34.42197.19––tg1282026-02-25community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)34.34422.67––tg1282025-06-13community–source
M2 Ultrallama 7B Q4_0llama.cpp (Vulkan)34.34205.98––tg1282025-11-25community–source
M3 Max (30 GPU)llama 7B Q8_0llama.cpp (Metal)34.3566.4–––2023-11-22community–source
GeForce RTX 5060 Tillama 7B Q4_0llama.cpp (Vulkan)34.19188.88––tg1282026-05-08communitycontradictionsource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)34.16180.97––tg1282025-06-18community–source
M1 Ultra (48 GPU)llama 7B F16llama.cpp (Metal)33.92875.81–––2023-11-22community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)33.9186.24––tg1282025-06-18community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)33.45161.47––tg1282025-08-22community–source
T4llama 7B Q4_0llama.cpp (Vulkan)33.36708.19––tg1282025-07-12community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)33.19182.18––tg1282025-11-26community–source
Radeon RX 9070 XTllama 7B Q4_0llama.cpp (ROCm)32.781728.29––tg1282026-04-01community–source
M5 (10 GPU)llama 7B Q4_0llama.cpp (Metal)31.88722.79–––2023-11-22community–source
M5llama 7B Q4_0llama.cpp (Metal)31.88722.79––tg1282026-08-30community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)31.74285.34––tg1282025-08-17communitycontradictionsource
M4 Max (40 GPU)llama 7B F16llama.cpp (Metal)31.64922.83–––2023-11-22community–source
M3 Prollama 7B Q4_0llama.cpp (Metal)31.54371.72––tg1282025-02-11community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)31.53102.55––tg1282025-10-27communitycontradictionsource
M3 Prollama 7B Q4_0llama.cpp (Metal)31.52368.92––tg1282025-02-11community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)31.45934.47––tg1282025-07-30community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)31.44651.09––tg1282025-05-22community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)31.39356.09––tg1282025-05-22community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)31.31468.82––tg1282025-08-17communitycontradictionsource
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)31.23103.24––tg1282025-10-27communitycontradictionsource
M5llama 7B Q4_0llama.cpp (Metal)31.02636.36––tg1282025-10-30community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)30.89226.82––tg1282025-05-22community–source
Arc B570llama 7B Q4_0llama.cpp (Vulkan)30.88342.71––tg1282025-05-22community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)30.77147.84––tg1282025-01-09community–source
M3 Pro (18 GPU)llama 7B Q4_0llama.cpp (Metal)30.74341.67–––2023-11-22community–source
M4 Pro (20 GPU)llama 7B Q8_0llama.cpp (Metal)30.69449.62–––2023-11-22community–source
M5llama 7B Q4_0llama.cpp (Metal)30.67650.65––tg1282025-10-30community–source
M3 Pro (14 GPU)llama 7B Q4_0llama.cpp (Metal)30.65269.49–––2023-11-22community–source
M4 Pro (16 GPU)llama 7B Q8_0llama.cpp (Metal)30.54367.13–––2023-11-22community–source
NVIDIA RTX 3500llama 7B Q4_0llama.cpp (CUDA)30.231406.43––tg1282025-09-20community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)29.52155.22––tg1282025-08-22community–source
Ryzen AI Max+ 395llama 7B Q4_0CPU29.52340.11––tg1282026-10-02community–source
Radeon RX 470 Gllama 7B Q4_0llama.cpp (Vulkan)29.28140.52––tg1282025-08-22community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (CPU)29.28304.53––tg1282025-11-12community–source
NVIDIA RTX 3500llama 7B Q4_0llama.cpp (CUDA)28.751610.14––tg1282025-09-20community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)28.37142.04––tg1282026-02-03communitycontradictionsource
Radeon RX 6500 XTllama 7B Q4_0llama.cpp (Vulkan)27.81255.25––tg1282025-05-04community–source
Ryzen AI Max+ 395llama 7B Q4_0CPU27.68300.27––tg1282026-10-02community–source
Radeon 8060Sllama 7B Q4_0llama.cpp (CPU)27.58308.17––tg1282025-11-12community–source
Arc A750llama 7B Q4_0llama.cpp (Vulkan)27.5788.86––tg1282025-01-08community–source
Radeon RX 580 Sllama 7B Q4_0llama.cpp (Vulkan)27.51212.32––tg1282025-06-02community–source
Radeon RX 480 Gllama 7B Q4_0llama.cpp (Vulkan)27.42237.27––tg1282025-09-21community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)26.57220.1––tg1282026-06-05community–source
M3 Prollama 7B Q4_0llama.cpp (Vulkan)26.39263.7––tg1282025-02-11community–source
M3 Prollama 7B Q4_0llama.cpp (Vulkan)26.23259.11––tg1282025-02-11community–source
M4llama 7B F16llama.cpp (Metal)26.05790.33––tg1282025-05-20community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)25.81213.39––tg1282026-06-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)25.63358.39––tg1282026-06-05community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)25.61277.2––tg1282025-06-13community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)25.57362.69––tg1282026-06-05community–source
Arc(TM) 140Vllama 7B Q4_0llama.cpp (Vulkan)25.12867.27––tg1282026-09-13community–source
M3 Max (40 GPU)llama 7B F16llama.cpp (Metal)25.09779.17–––2023-11-22community–source
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)24.76435.39––tg1282025-07-17community–source
M2 Max (38 GPU)llama 7B F16llama.cpp (Metal)24.65755.67–––2023-11-22community–source
Radeon RX 570 Sllama 7B Q4_0llama.cpp (Vulkan)24.6353.68––tg1282026-04-25communitycontradictionsource
Radeon RX 6800Mllama 7B Q4_0llama.cpp (Vulkan)24.56480.52––tg1282025-07-17community–source
Arc(TM) 140Vllama 7B Q4_0llama.cpp (Vulkan)24.37864.99––tg1282026-09-13community–source
M4 Max (32 GPU)llama 7B F16llama.cpp (Metal)24.29736.25–––2023-11-22community–source
M2 Max (30 GPU)llama 7B F16llama.cpp (Metal)24.16600.46–––2023-11-22community–source
M4 (10 GPU)llama 7B Q4_0llama.cpp (Metal)24.11221.29–––2023-11-22community–source
Radeon RX 570 Sllama 7B Q4_0llama.cpp (Vulkan)24.1152.66––tg1282026-04-25communitycontradictionsource
GB10llama 7B Q4_0llama.cpp (CPU)24.08313.01––tg1282026-01-20community–source
M2 Maxllama 7B Q4_0llama.cpp (Vulkan)24.0779.94––tg1282024-06-21community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)24.04224.16––tg1282026-06-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)23.97491.34––tg1282026-06-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)23.84495.16––tg1282026-06-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)23.81226.11––tg1282026-06-05community–source
A4000llama 7B Q4_0llama.cpp (CUDA)23.76311.55––tg1282025-08-19community–source
A4000llama 7B Q4_0llama.cpp (CUDA)23.63309.3––tg1282025-08-19community–source
Arc A750llama 7B Q4_0llama.cpp (SYCL)23.61506.22––tg1282025-11-07community–source
M4llama 7B Q4_0llama.cpp (Metal)23.29212.13––tg1282024-05-20community–source
Radeon RX 5700 XTllama 7B Q4_0llama.cpp (Vulkan)23.21214.48––tg1282025-01-12communitycontradictionsource
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)23.1373.78––tg1282025-12-14community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)23.06374.88––tg1282025-12-14community–source
M1 Max (32 GPU)llama 7B F16llama.cpp (Metal)23.03599.53–––2023-11-22community–source
M2 Pro (19 GPU)llama 7B Q8_0llama.cpp (Metal)23.01344.5–––2023-11-22community–source
M2 Maxllama 7B Q4_0llama.cpp (CPU)22.8649.04––tg1282024-06-21community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)22.8307.71––tg1282025-06-18community–source
M2 Pro (16 GPU)llama 7B Q8_0llama.cpp (Metal)22.7288.46–––2023-11-22community–source
M1 Max (24 GPU)llama 7B F16llama.cpp (Metal)22.55453.03–––2023-11-22community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)22.41311.56––tg1282025-06-18community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)22.37545.1––tg1282025-12-14community–source
M1 Pro (16 GPU)llama 7B Q8_0llama.cpp (Metal)22.34270.37–––2023-11-22community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)22.03318.6––tg1282025-12-14community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)21.99329.76––tg1282025-12-14community–source
M1 Pro (14 GPU)llama 7B Q8_0llama.cpp (Metal)21.95235.16–––2023-11-22community–source
M2 (10 GPU)llama 7B Q4_0llama.cpp (Metal)21.91179.57–––2023-11-22community–source
M2 (8 GPU)llama 7B Q4_0llama.cpp (Metal)21.7145.91–––2023-11-22community–source
M5 Pro (20 GPU)llama 7B F16llama.cpp (Metal)21.551588.78–––2023-11-22community–source
M5 Prollama 7B F16llama.cpp (Metal)21.551588.78––tg1282026-08-31community–source
M3llama 7B Q4_0llama.cpp (ROCm)21.47310.95––tg1282024-01-21community–source
M5 Prollama 7B F16llama.cpp (Metal)21.391310.78––tg1282026-09-24community–source
M5 Prollama 7B F16llama.cpp (Metal)21.37695.51––tg1282026-04-14community–source
M3 (10 GPU)llama 7B Q4_0llama.cpp (Metal)21.34186.75–––2023-11-22community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)21.28311.02––tg1282026-05-01community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)21.26240.89––tg1282025-10-19community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)21.18424.24––tg1282025-12-14community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)21.15277.91––tg1282025-10-19community–source
M5 Prollama 7B F16llama.cpp (Metal)21.111310.28––tg1282026-04-14community–source
M6llama 7B 6.74BQ8_0llama.cpp (Metal)21.01970.46––tg1282026-10-01community–source
M2 Prollama 7B Q4_0llama.cpp (Vulkan)20.9758.86––tg1282025-08-17community–source
M2 Prollama 7B Q4_0llama.cpp (Vulkan)20.9562.736.5–tg1282025-08-17community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)20.59297.39––tg1282025-11-04community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)20.53266.19––tg1282025-11-04community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)20.26227.06––tg1282026-06-13community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)20.18254.09––tg1282025-03-20community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)20.1245.79––tg1282025-02-09community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)20.0291.42––tg1282026-01-17community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.93257.01––tg1282025-08-14community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)19.91281.62––tg1282025-03-17community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)19.81267.27––tg1282026-05-01community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)19.81349.57––tg1282025-11-04community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.81264.67––tg1282025-08-14community–source
M2 Maxllama 7B Q8_0llama.cpp (Vulkan)19.7979.79––tg1282024-06-21community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.78314.97––tg1282026-03-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)19.76441.14––tg1282026-06-13community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)19.7216.06––tg1282026-06-05community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)19.66326.84––tg1282025-11-04community–source
M3 Max (30 GPU)llama 7B F16llama.cpp (Metal)19.54589.41–––2023-11-22community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.39354.86––tg1282025-09-21community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.29380.03––tg1282025-09-21community–source
Arc A380llama 7B Q4_0llama.cpp (OpenVINO)19.211513.67––tg1282026-06-13community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)19.16278.4934–tg1282026-01-17community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)19.12351.23––tg1282026-06-05community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)19.08110.39––tg1282026-03-05community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)19.06531.03––tg1282026-01-17community–source
Radeon 840Mllama 7B Q4_0llama.cpp (Vulkan)19.03150.12––tg1282026-08-07community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)18.92102.86––tg1282026-06-21community–source
Radeon 760Mllama 7B Q4_0llama.cpp (Vulkan)18.89219.24––tg1282026-04-25community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)18.76287.18––tg1282026-03-05community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)18.62351.23––tg1282026-06-13community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)18.54203.17––tg1282026-06-13community–source
M5 (10 GPU)llama 7B Q8_0llama.cpp (Metal)18.42715.42–––2023-11-22community–source
M5llama 7B Q8_0llama.cpp (Metal)18.42715.42––tg1282026-08-30community–source
AMD Custom GPU 0932llama 7B 6.74BQ4_0llama.cpp (Vulkan)18.40164.34––tg1282026-08-20community–source
RTX 3080llama 7B Q4_0llama.cpp (Vulkan)18.32230.79––tg1282026-04-03community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)18.24121.84––tg1282026-03-05community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)18.22237.73––tg1282025-03-01community–source
Arc A770llama 7B Q4_0llama.cpp (SYCL)18.22288.17––tg1282025-07-30community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)18.19478.53––tg1282026-06-05community–source
Arc A380llama 7B Q4_0llama.cpp (OpenVINO)18.181497.21––tg1282026-06-12community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)18.16220.98––tg1282026-06-05community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)18.02437.1––tg1282026-01-17community–source
Arc A380llama 7B Q4_0llama.cpp (Vulkan)17.87263.7––tg1282026-06-13community–source
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)17.84218.32––tg1282026-05-25communitycontradictionsource
Radeon 760Mllama 7B Q4_0llama.cpp (Vulkan)17.69187.02––tg1282026-04-25community–source
Radeon 840Mllama 7B Q4_0llama.cpp (Vulkan)17.66131.33––tg1282026-08-07community–source
Arc A380llama 7B Q4_0llama.cpp (Vulkan)17.57259.51––tg1282026-06-12community–source
M5llama 7B Q8_0llama.cpp (Metal)17.55512.15––tg1282025-10-30community–source
Arc A380llama 7B Q4_0llama.cpp (Vulkan)17.55249.81––tg1282026-06-13community–source
M3 Pro (18 GPU)llama 7B Q8_0llama.cpp (Metal)17.53344.66–––2023-11-22community–source
AMD Custom GPU 0932llama 7B 6.74BQ4_0llama.cpp (Vulkan)17.52144.31––tg1282026-08-20community–source
M5llama 7B Q8_0llama.cpp (Metal)17.5489.78––tg1282025-10-30community–source
M3 Pro (14 GPU)llama 7B Q8_0llama.cpp (Metal)17.44272.11–––2023-11-22community–source
Arc A380llama 7B Q4_0llama.cpp (Vulkan)17.25247.03––tg1282026-06-12community–source
M4 Pro (16 GPU)llama 7B F16llama.cpp (Metal)17.19381.14–––2023-11-22community–source
M4 Pro (20 GPU)llama 7B F16llama.cpp (Metal)17.18464.48–––2023-11-22community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)17.096.83––tg1282026-06-21community–source
Quadro RTX 5000llama 7B Q4_0llama.cpp (CUDA)16.74216.34––tg1282026-05-25communitycontradictionsource
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)16.73421.3––tg1282026-05-13community–source
Radeon 660Mllama 7B Q4_0llama.cpp (Vulkan)16.64138.61––tg1282025-10-19community–source
Radeon 780Mllama 7B Q4_0llama.cpp (Vulkan)16.6205.15––tg1282025-03-17communitycontradictionsource
A770llama 7B Q4_0llama.cpp (SYCL)16.57437.75––tg1282026-01-28communitycontradictionsource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)16.5675.59––tg1282025-06-18communitycontradictionsource
Radeon 660Mllama 7B Q4_0llama.cpp (Vulkan)16.56148.05––tg1282025-10-19community–source
Radeon 660Mllama 7B Q4_0llama.cpp (Vulkan)16.2136.56––tg1282025-09-09community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)16.17276.43––tg1282026-06-12community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)16.16213.31––tg1282025-10-28community–source
Radeon 840Mllama 7B Q4_0llama.cpp (ROCm)16.12172.58––tg1282026-08-07community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)16.12417.37––tg1282026-05-13community–source
Radeon 660Mllama 7B Q4_0llama.cpp (Vulkan)16.11147.9––tg1282025-09-09community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)16.08100.35––tg1282025-08-27community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)16.0649.49––tg1282026-02-15communitycontradiction outliersource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)16.0578.79––tg1282025-06-18communitycontradictionsource
Arc A380llama 7B Q4_0llama.cpp (SYCL)15.97170.14––tg1282026-06-12community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)15.42325.54––tg1282026-10-10community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)15.11246.65––tg1282025-10-27community–source
Radeon RX 9060 XTllama 7B Q4_0llama.cpp (Vulkan)15.07103.01––tg1282026-02-03communitycontradictionsource
Arc A380llama 7B Q4_0llama.cpp (SYCL)14.96201.62––tg1282026-06-12community–source
Arc A380llama 7B Q4_0llama.cpp (SYCL)14.95347.28––tg1282026-06-12community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)14.94201.59––tg1282025-10-27community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)14.71494.78––tg1282026-02-19community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)14.58260.18––tg1282026-10-10community–source
Radeon 840Mllama 7B Q4_0llama.cpp (ROCm)14.33134.09––tg1282026-08-07community–source
M1 (7 GPU)llama 7B Q4_0llama.cpp (Metal)14.19107.81–––2023-11-22community–source
M1 (8 GPU)llama 7B Q4_0llama.cpp (Metal)14.15117.96–––2023-11-22community–source
M1llama 7B Q4_0llama.cpp (Metal)13.99131.46––tg1282025-05-20community–source
Radeon RX 580llama 7B Q4_0llama.cpp (Vulkan)13.9985.37––tg1282025-08-27community–source
Radeon 680Mllama 7B Q4_0llama.cpp (Vulkan)13.86670.41––tg1282026-02-19community–source
M2 Maxllama 7B Q8_0llama.cpp (CPU)13.7952.83––tg1282024-06-21community–source
M4 (10 GPU)llama 7B Q8_0llama.cpp (Metal)13.54223.64–––2023-11-22community–source
M2llama 7B Q4_0llama.cpp (Vulkan)13.550.79––tg1282025-11-07community–source
A770llama 7B Q4_0llama.cpp (SYCL)13.23480.14––tg1282025-09-21communitycontradictionsource
Radeon 840Mllama 7B Q4_0llama.cpp (CPU)13.1861.42––tg1282026-08-07community–source
M2 Pro (19 GPU)llama 7B F16llama.cpp (Metal)13.06384.38–––2023-11-22community–source
Radeon 890Mllama 7B Q4_0llama.cpp (ROCm)12.97163.25––tg1282025-10-28community–source
Radeon 840Mllama 7B Q4_0llama.cpp (CPU)12.9560.33––tg1282026-08-07community–source
M1llama 7B Q4_0llama.cpp (Vulkan)12.8535.93––tg1282025-11-21community–source
M1 Pro (14 GPU)llama 7B F16llama.cpp (Metal)12.75262.65–––2023-11-22community–source
M1 Pro (16 GPU)llama 7B F16llama.cpp (Metal)12.75302.14–––2023-11-22community–source
Radeon RX 580 Sllama 7B Q4_0llama.cpp (Vulkan)12.72131.42––tg1282025-09-07community–source
Radeon 660Mllama 7B Q4_0llama.cpp (ROCm)12.72120.52––tg1282025-12-10community–source
M2 Pro (16 GPU)llama 7B F16llama.cpp (Metal)12.47312.65–––2023-11-22community–source
M1llama 7B Q4_0llama.cpp (Vulkan)12.4738.29––tg1282025-11-21community–source
M1llama 7B Q4_0llama.cpp (Vulkan)12.4131.3120.35–tg1282025-08-17community–source
M1llama 7B Q4_0llama.cpp (Vulkan)12.3828.65––tg1282025-08-17community–source
M3 (10 GPU)llama 7B Q8_0llama.cpp (Metal)12.27187.52–––2023-11-22community–source
M2llama 7B Q4_0llama.cpp (Vulkan)12.2546.81––tg1282025-11-07community–source
M2 (10 GPU)llama 7B Q8_0llama.cpp (Metal)12.21181.4–––2023-11-22community–source
M2 (8 GPU)llama 7B Q8_0llama.cpp (Metal)12.18147.27–––2023-11-22community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (Vulkan)12.1626.05––tg1282025-03-10community–source
Radeon RX 6600llama 7B Q4_0llama.cpp (CPU)12.0826.26––tg1282025-03-10community–source
RTX 3060llama 7B Q4_K - Mediumllama.cpp (Vulkan)11.8218.01––tg1282025-11-18community–source
Radeon RX 580 Sllama 7B Q4_0llama.cpp (Vulkan)11.62177.33––tg1282025-09-07community–source
M6llama 7B 6.74BF16llama.cpp (Metal)11.59952.53––tg1282026-10-01community–source
A770llama 7B Q4_0llama.cpp (SYCL)11.14497.1––tg1282026-01-28communitycontradictionsource
M2llama 7B Q4_0llama.cpp (Vulkan)11.0738.67––tg1282025-01-08community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)10.8190.15––tg1282025-03-20communitycontradictionsource
Radeon RX 550llama 7B Q4_0llama.cpp (Vulkan)10.252.66––tg1282025-11-16community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)10.0976.53––tg1282025-06-18communitycontradictionsource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)10.0681.59––tg1282025-06-18communitycontradiction outliersource
M3 Pro (18 GPU)llama 7B F16llama.cpp (Metal)9.89357.45–––2023-11-22community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)9.87113.32––tg1282026-02-12communitycontradiction outliersource
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)9.7972.76––tg1282025-12-03community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (CPU)9.7244.12––tg1282026-07-14community–source
Radeon RX 7600llama 7B Q4_0llama.cpp (CPU)9.7242.94––tg1282026-07-14community–source
M5llama 7B F16llama.cpp (Metal)9.67374.35––tg1282025-10-30community–source
Radeon RX 6800llama 7B Q4_0llama.cpp (Vulkan)9.63103.87––tg1282026-02-12communitycontradiction outliersource
M5llama 7B F16llama.cpp (Metal)9.56378.36––tg1282025-10-30community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)9.4874.29––tg1282025-12-03community–source
RTX 3060llama 7B Q4_K - Mediumllama.cpp (Vulkan)9.41081.37––tg1282025-11-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)8.55117.48––tg1282025-06-18communitycontradictionsource
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)8.31293.41––tg1282025-10-27community–source
Radeon 890Mllama 7B Q4_0llama.cpp (Vulkan)8.3315.24––tg1282025-10-27communitycontradictionsource
M2 Maxllama 7B F16llama.cpp (CPU)8.1926.18––tg1282024-06-21community–source
Radeon RX 7600 XTllama 7B Q4_0llama.cpp (Vulkan)8.1734.95––tg1282026-02-15communitycontradiction outliersource
M1 (7 GPU)llama 7B Q8_0llama.cpp (Metal)7.92108.21–––2023-11-22community–source
M1 (8 GPU)llama 7B Q8_0llama.cpp (Metal)7.91117.25–––2023-11-22community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)7.6991.04––tg1282025-12-03community–source
M1llama 7B Q8_0llama.cpp (Metal)7.67133.34––tg1282025-05-20community–source
M4 (10 GPU)llama 7B F16llama.cpp (Metal)7.43230.18–––2023-11-22community–source
RTX 3060llama 7B Q4_0llama.cpp (Vulkan)7.3390.13––tg1282025-12-03community–source
M2 (10 GPU)llama 7B F16llama.cpp (Metal)6.72201.34–––2023-11-22community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.8216.62––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.7117.7––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.6142.12––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.5441.26––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.4215.22––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)5.3519.18––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)2.8784.19––tg1282025-06-18communitycontradictionsource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)2.85107.8––tg1282025-06-18communitycontradictionsource
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)2.6723.19––tg1282025-06-18community–source
RTX 3090llama 7B Q4_0llama.cpp (Vulkan)2.6323.19––tg1282025-06-18community–source
M2 Maxllama 7B F16llama.cpp (Vulkan)2.0280.4––tg1282024-06-21community–source

← all models