llama 7B mostly

158 records across 21 hardware.

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
M5 Maxllama 7B mostly Q4_0llama.cpp (Metal)104.03990.53––tg1282026-03-31community–source
M5 Maxllama 7B mostly Q4_0llama.cpp (Metal)102.93987.1––tg1282026-03-09community–source
M2 Ultrallama 7B mostly Q4_0llama.cpp (Metal)95.341240.61––tg1282023-12-04community–source
M2 Ultrallama 7B mostly Q4_0llama.cpp (Metal)94.271238.48––tg1282023-11-22community–source
M3 Ultra 24+8llama 7B mostly Q4_0llama.cpp (Metal)92.141471.24––tg1282025-03-23community–source
M3 Ultrallama 7B mostly Q4_0llama.cpp (Metal)91.931473.76––tg1282025-04-22community–source
M2 Ultrallama 7B mostly Q4_0llama.cpp (Metal)88.641013.81––tg1282023-11-23community–source
M3 Ultra 20+8llama 7B mostly Q4_0llama.cpp (Metal)88.41073.09––tg1282025-03-14community–source
M3 Max 10+4llama 7B mostly Q4_0llama.cpp (Metal)83.731030.04––tg1282024-01-11community–source
M4 Maxllama 7B mostly Q4_0llama.cpp (Metal)83.06885.68––tg1282024-12-04community–source
M4 Maxllama 7B mostly Q4_0llama.cpp (Metal)82.52746.09––tg1282024-12-17community–source
M4 Maxllama 7B mostly Q4_0llama.cpp (Metal)82.36884.59––tg1282025-02-19community–source
M1 Ultrallama 7B mostly Q4_0llama.cpp (Metal)74.93772.24––tg1282023-12-01community–source
M3llama 7B mostly Q4_0llama.cpp (CUDA)73.381816.45––tg1282024-01-21community–source
M3llama 7B mostly Q4_0llama.cpp (CUDA)69.831956.45––tg1282024-01-21community–source
M3llama 7B mostly Q4_0llama.cpp (CUDA)69.491087.59––tg1282024-01-11community–source
M4llama 7B mostly Q4_0llama.cpp (Metal)69.24715.74––tg1282025-05-20community–source
M2 Ultrallama 7B mostly Q8_0llama.cpp (Metal)67.361250.86––tg1282023-12-04community–source
M2 Ultrallama 7B mostly Q8_0llama.cpp (Metal)66.641248.59––tg1282023-11-22community–source
M3 Maxllama 7B mostly Q4_0llama.cpp (Metal)66.31759.7––tg1282023-11-24community–source
M2 Maxllama 7B mostly Q4_0llama.cpp (Metal)65.95671.31––tg1282023-11-22community–source
M3 Maxllama 7B mostly Q4_0llama.cpp (Metal)65.85690.99––tg1282023-11-22community–source
M5 Maxllama 7B mostly Q8_0llama.cpp (Metal)65.111054.83––tg1282026-03-31community–source
M5 Maxllama 7B mostly Q8_0llama.cpp (Metal)64.611051.59––tg1282026-03-09community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)64.55669.51––tg1282023-11-25community–source
M3 Ultrallama 7B mostly Q8_0llama.cpp (Metal)64.161488.84––tg1282025-04-22community–source
M3 Ultra 24+8llama 7B mostly Q8_0llama.cpp (Metal)63.931487.51––tg1282025-03-23community–source
M3 Ultra 20+8llama 7B mostly Q8_0llama.cpp (Metal)63.551085.76––tg1282025-03-14community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)63.29664.07––tg1282023-11-25community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)62.98668.28––tg1282023-11-26community–source
M2 Ultrallama 7B mostly Q8_0llama.cpp (Metal)62.141003.16––tg1282023-11-23community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (CUDA)61.391945.87––tg1282023-12-17community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (Metal)61.19530.06––tg1282023-11-25community–source
M2 Maxllama 7B mostly Q4_0llama.cpp (Metal)60.99537.6––tg1282023-11-29community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (Metal)60.96529.4––tg1282023-12-17community–source
M1llama 7B mostly Q4_0llama.cpp (Metal)60.4527.76––tg1282023-12-21community–source
M5 Prollama 7B mostly Q4_0llama.cpp (Metal)60.04403.19––tg1282026-04-14community–source
M3 Max 10+4llama 7B mostly Q8_0llama.cpp (Metal)59.871042.95––tg1282024-01-11community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (CUDA)57.871183.28––tg10242023-12-17community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (Metal)57.39466.05––tg10242023-12-17community–source
M3 Max 10+4llama 7B mostly Q4_0llama.cpp (Metal)56.85567.59––tg1282023-12-22community–source
M1 Ultrallama 7B mostly Q8_0llama.cpp (Metal)55.69783.45––tg1282023-12-01community–source
M1 Maxllama 7B mostly Q4_0llama.cpp (Metal)54.61400.26––tg1282023-12-02communitycontradictionsource
M4 Maxllama 7B mostly Q8_0llama.cpp (Metal)54.05891.94––tg1282024-12-04community–source
M4 Maxllama 7B mostly Q8_0llama.cpp (Metal)53.75891.32––tg1282025-02-19community–source
M4 Maxllama 7B mostly Q8_0llama.cpp (Metal)53.06852.47––tg1282024-12-17community–source
M3llama 7B mostly Q8_0llama.cpp (CUDA)50.751937.77––tg1282024-01-21community–source
M4 Prollama 7B mostly Q4_0llama.cpp (Metal)50.74439.78––tg1282024-11-08community–source
M3llama 7B mostly Q8_0llama.cpp (CUDA)50.471049.96––tg1282024-01-11community–source
M4 Prollama 7B mostly Q4_0llama.cpp (Metal)49.64364.06––tg1282024-12-07community–source
M4llama 7B mostly Q8_0llama.cpp (Metal)43.8720.38––tg1282025-05-20community–source
M3 Maxllama 7B mostly Q8_0llama.cpp (Metal)43.0749.37––tg1282023-11-22community–source
M3 Maxllama 7B mostly Q8_0llama.cpp (Metal)42.75757.64––tg1282023-11-24community–source
M3llama 7B mostly Q8_0llama.cpp (CUDA)42.431688.37––tg1282024-01-21community–source
M3 Maxllama 7B mostly Q8_0llama.cpp (Metal)42.37707.07––tg1282023-12-28community–source
M3 Ultra 20+8llama 7B mostly F16llama.cpp (Metal)42.241121.8––tg1282025-03-14community–source
M2 Maxllama 7B mostly Q8_0llama.cpp (Metal)41.83677.91––tg1282023-11-22community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)41.79674.5––tg1282023-11-25community–source
M2 Ultrallama 7B mostly F16llama.cpp (Metal)41.341408.01––tg1282023-12-04community–source
M2 Ultrallama 7B mostly F16llama.cpp (Metal)41.021401.85––tg1282023-11-22community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)40.85671.33––tg1282023-11-25community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)40.67674.37––tg1282023-11-26community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (Metal)40.2537.37––tg1282023-11-25community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (Metal)40.17537.23––tg1282023-12-17community–source
M3 Ultrallama 7B mostly F16llama.cpp (Metal)40.11527.74––tg1282025-04-22community–source
M2 Maxllama 7B mostly Q8_0llama.cpp (Metal)39.97540.15––tg1282023-11-29community–source
M2 Ultrallama 7B mostly F16llama.cpp (Metal)39.861128.59––tg1282023-11-23community–source
M3 Ultra 24+8llama 7B mostly F16llama.cpp (Metal)39.781538.34––tg1282025-03-23community–source
M2 Prollama 7B mostly Q4_0llama.cpp (Metal)39.71342.61––tg1282023-12-28community–source
M1llama 7B mostly Q8_0llama.cpp (Metal)39.6536.5––tg1282023-12-21community–source
M2 Prollama 7B mostly Q4_0llama.cpp (Metal)38.86341.19––tg1282023-12-11community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (Metal)38.5471.64––tg10242023-12-17community–source
M5 Maxllama 7B mostly F16llama.cpp (Metal)38.061032.18––tg1282026-03-31community–source
M2 Prollama 7B mostly Q4_0llama.cpp (Metal)37.87294.24––tg1282023-11-24community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (Metal)37.81405.87––tg1282023-12-02community–source
M5 Maxllama 7B mostly F16llama.cpp (Metal)37.581018.3––tg1282026-03-09community–source
M3 Max 10+4llama 7B mostly F16llama.cpp (Metal)37.011168.89––tg1282024-01-11community–source
M1 Prollama 7B mostly Q4_0llama.cpp (Metal)36.41266.25––tg1282023-11-22community–source
M1 Prollama 7B mostly Q4_0llama.cpp (Metal)36.34233.49––tg1282024-01-02community–source
M1 Ultrallama 7B mostly F16llama.cpp (Metal)36.311158.95––tg1282024-01-10community–source
M5 Prollama 7B mostly Q8_0llama.cpp (Metal)35.86431.14––tg1282026-04-14community–source
M1 Prollama 7B mostly Q4_0llama.cpp (Metal)35.45233.01––tg1282023-12-17community–source
M3llama 7B mostly F16llama.cpp (CUDA)35.42925.62––tg1282024-01-21community–source
M3llama 7B mostly F16llama.cpp (CUDA)35.28962.94––tg1282024-01-11community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (CUDA)35.021873.64––tg1282023-12-17community–source
M3 Max 10+4llama 7B mostly Q8_0llama.cpp (Metal)34.3566.4––tg1282023-12-22community–source
M3 Max 10+4llama 7B mostly Q8_0llama.cpp (Metal)34.3566.4––tg1282024-01-11communitycontradictionsource
M1 Ultrallama 7B mostly F16llama.cpp (Metal)33.92875.81––tg1282023-12-01community–source
M1 Maxllama 7B mostly Q8_0llama.cpp (CUDA)33.861157.46––tg10242023-12-17community–source
M4 Maxllama 7B mostly F16llama.cpp (Metal)31.64922.83––tg1282024-12-04community–source
M4 Maxllama 7B mostly F16llama.cpp (Metal)31.61923.55––tg1282024-12-17community–source
M4 Maxllama 7B mostly F16llama.cpp (Metal)31.56920.48––tg1282025-02-19community–source
M3 Prollama 7B mostly Q4_0llama.cpp (Metal)30.74341.67––tg1282023-12-09community–source
M4 Prollama 7B mostly Q8_0llama.cpp (Metal)30.69449.62––tg1282024-11-08community–source
M3 Prollama 7B mostly Q4_0llama.cpp (Metal)30.65269.49––tg1282023-11-25community–source
M1 Prollama 7B mostly Q4_0llama.cpp (Metal)30.6205.41––tg1282023-12-17community–source
M4 Prollama 7B mostly Q8_0llama.cpp (Metal)30.54367.13––tg1282024-12-07community–source
M3llama 7B mostly Q4_0llama.cpp (Metal)30.36338.37––tg1282023-12-01community–source
M5llama 7B mostly Q4_0llama.cpp (Metal)29.62247.68––tg1282025-10-30community–source
M4llama 7B mostly F16llama.cpp (Metal)25.58747.59––tg1282025-05-20community–source
M3 Maxllama 7B mostly F16llama.cpp (Metal)25.27794.26––tg1282023-11-22community–source
M3 Maxllama 7B mostly F16llama.cpp (Metal)25.14784.83––tg1282023-12-28community–source
M3 Maxllama 7B mostly F16llama.cpp (Metal)25.09779.17––tg1282023-11-24community–source
M2 Maxllama 7B mostly F16llama.cpp (Metal)24.65755.67––tg1282023-11-22community–source
M2llama 7B mostly F16llama.cpp (Metal)24.54747.99––tg1282023-11-26community–source
M2llama 7B mostly F16llama.cpp (Metal)24.31754.39––tg1282023-11-25community–source
M2 Maxllama 7B mostly F16llama.cpp (Metal)24.16600.46––tg1282023-11-29community–source
M4llama 7B mostly Q4_0llama.cpp (Metal)24.11221.29––tg1282024-11-10community–source
M3llama 7B mostly F16llama.cpp (CUDA)23.933022.37––tg1282024-01-21community–source
M2 Prollama 7B mostly Q8_0llama.cpp (Metal)23.24335.9––tg1282023-12-28community–source
M1 Maxllama 7B mostly F16llama.cpp (Metal)23.03599.53––tg1282023-11-25community–source
M2 Prollama 7B mostly Q8_0llama.cpp (Metal)23.01344.5––tg1282023-12-11community–source
M1 Maxllama 7B mostly F16llama.cpp (Metal)22.96596.85––tg1282023-12-17community–source
M1llama 7B mostly F16llama.cpp (Metal)22.83598.96––tg1282023-12-21community–source
M2 Prollama 7B mostly Q8_0llama.cpp (Metal)22.7288.46––tg1282023-11-24community–source
M1 Maxllama 7B mostly F16llama.cpp (Metal)22.55453.03––tg1282023-12-02community–source
M1 Prollama 7B mostly Q8_0llama.cpp (Metal)22.51237.0––tg1282024-01-02community–source
M1 Maxllama 7B mostly F16llama.cpp (Metal)22.41516.68––tg10242023-12-17community–source
M1 Prollama 7B mostly Q8_0llama.cpp (Metal)22.34270.37––tg1282023-11-22community–source
M1 Prollama 7B mostly Q8_0llama.cpp (Metal)22.16236.72––tg1282023-12-17community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)21.91179.57––tg1282023-11-22community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)21.7145.91––tg1282024-01-07community–source
M3llama 7B mostly Q4_0llama.cpp (Metal)21.34186.75––tg1282023-11-27community–source
M1 Prollama 7B mostly Q8_0llama.cpp (Metal)20.55201.58––tg1282023-12-17community–source
M5 Prollama 7B mostly F16llama.cpp (Metal)20.22423.3––tg1282026-04-14community–source
M3 Max 10+4llama 7B mostly F16llama.cpp (Metal)19.54589.41––tg1282023-12-22community–source
M1 Maxllama 7B mostly F16llama.cpp (CUDA)19.452757.29––tg1282023-12-17community–source
M3llama 7B mostly Q4_0llama.cpp (CUDA)19.23671.1––tg1282024-01-11communitycontradiction outliersource
M3llama 7B mostly Q8_0llama.cpp (CUDA)19.11637.51––tg1282024-01-11communitycontradictionsource
M1 Maxllama 7B mostly F16llama.cpp (CUDA)19.11453.27––tg10242023-12-17community–source
M3llama 7B mostly F16llama.cpp (CUDA)18.46648.07––tg1282024-01-11communitycontradictionsource
M3 Prollama 7B mostly Q8_0llama.cpp (Metal)17.53344.66––tg1282023-12-09community–source
M3 Prollama 7B mostly Q8_0llama.cpp (Metal)17.44272.11––tg1282023-11-25community–source
M4 Prollama 7B mostly F16llama.cpp (Metal)17.19381.14––tg1282024-12-07community–source
M4 Prollama 7B mostly F16llama.cpp (Metal)17.18464.48––tg1282024-11-08community–source
M5llama 7B mostly Q8_0llama.cpp (Metal)16.62264.15––tg1282025-10-30community–source
M3llama 7B mostly Q8_0llama.cpp (Metal)16.46339.31––tg1282023-12-01community–source
M1llama 7B mostly Q4_0llama.cpp (Metal)14.19107.81––tg1282023-12-15community–source
M1llama 7B mostly Q4_0llama.cpp (Metal)14.15117.96––tg1282023-12-02community–source
M1llama 7B mostly Q4_0llama.cpp (Metal)14.13115.67––tg1282025-05-20community–source
M4llama 7B mostly Q8_0llama.cpp (Metal)13.54223.64––tg1282024-11-10community–source
M2 Prollama 7B mostly F16llama.cpp (Metal)13.06384.38––tg1282023-12-11community–source
M2 Prollama 7B mostly F16llama.cpp (Metal)12.97373.97––tg1282023-12-28community–source
M1 Prollama 7B mostly F16llama.cpp (Metal)12.75302.14––tg1282023-11-22community–source
M1 Prollama 7B mostly F16llama.cpp (Metal)12.75262.65––tg1282024-01-02community–source
M2 Prollama 7B mostly F16llama.cpp (Metal)12.47312.65––tg1282023-11-24community–source
M3llama 7B mostly Q8_0llama.cpp (Metal)12.27187.52––tg1282023-11-27community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)12.21181.4––tg1282023-11-22community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)12.18147.27––tg1282024-01-07community–source
M3 Prollama 7B mostly F16llama.cpp (Metal)9.89357.45––tg1282023-12-09community–source
M5llama 7B mostly F16llama.cpp (Metal)9.43257.79––tg1282025-10-30community–source
M2llama 7B mostly Q4_0llama.cpp (Metal)8.6185.24––tg1282023-12-05communitycontradictionsource
M1llama 7B mostly Q8_0llama.cpp (Metal)7.92108.21––tg1282023-12-15community–source
M1llama 7B mostly Q8_0llama.cpp (Metal)7.91117.25––tg1282023-12-02community–source
M1llama 7B mostly Q8_0llama.cpp (Metal)7.69121.73––tg1282025-05-20community–source
M4llama 7B mostly F16llama.cpp (Metal)7.43230.18––tg1282024-11-10community–source
M2llama 7B mostly F16llama.cpp (Metal)6.72201.34––tg1282023-11-22community–source
M2llama 7B mostly Q8_0llama.cpp (Metal)5.3176.9––tg1282023-12-05communitycontradictionsource

← all models