RTX 5090
152 records across 22 models (quants: APEX-I-MiniPlus V2.1, BF16, F16, Gyro-M, Gyro-S, MXFP4, MXFP4 (experts), NVFP4, Q4_0, Q4_K, Q4_K_M, Q4_K_S, Q4_blend2, Q4_blend4, Q4_blend5, Q6_K, Ternary PQ2_0, Ternary PTQ1_0).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RTX 5090 | DeepSeek-V4.1-Flash | MXFP4 (experts) | llama.cpp (JigSawPT DSV4.1 port) | 21.3 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | DeepSeek-V4.1-Flash | MXFP4 (experts) | llama.cpp (JigSawPT DSV4.1 port) | 21 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | DeepSeek-V4.1-Flash | MXFP4 (experts) | llama.cpp (JigSawPT DSV4.1 port) | 6.2 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | DeepSeek-V4.1-Flash | MXFP4 (experts) | llama.cpp (JigSawPT DSV4.1 port) | 5.1 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | Gemma 4 12B | Q4_K_M | llama.cpp | 279.09 | – | – | – | – | 2026-10-12 | sourced | – | source |
| RTX 5090 | Gemma 4 12B | Q4_K_M | llama.cpp | 278.37 | – | – | – | – | 2026-10-12 | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 180.3 | 8799.2 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 167.2 | 7733.2 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 159.4 | 6292.6 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 149.4 | 4360.6 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 130.2 | 2839.1 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | Gemma4 26B 26B | Q4_K | llama.cpp | 106.0 | 1707.2 | – | – | 262144 | – | sourced | – | source |
| RTX 5090 | Gemma4 31B 31B | Q4_K | llama.cpp | 61.1 | 3395.0 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Gemma4 31B 31B | Q4_K | llama.cpp | 59.2 | 2794.3 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Gemma4 31B 31B | Q4_K | llama.cpp | 55.4 | 2229.1 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Gemma4 31B 31B | Q4_K | llama.cpp | 51.6 | 1459.3 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | Gemma4 31B 31B | Q4_K | llama.cpp | 43.4 | 900.2 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | gpt-oss 20B 20B | Q4_K | llama.cpp | 298.2 | 9443.8 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | gpt-oss 20B 20B | Q4_K | llama.cpp | 249.2 | 7168.1 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | gpt-oss 20B 20B | Q4_K | llama.cpp | 215.1 | 5183.1 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | gpt-oss 20B 20B | Q4_K | llama.cpp | 169.0 | 3019.7 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | gpt-oss 20B 20B | Q4_K | llama.cpp | 112.0 | 1636.4 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | KAT-Coder-V2.5 | APEX-I-MiniPlus V2.1 | llama.cpp | 247 | 2800 | – | – | – | 2026-09-18 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 509.6 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 500.7 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 375.1 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 231.93 | 12761.21 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 230.70 | 12850.64 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 173.43 | 12352.17 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 162.95 | 12388.47 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 161.79 | 12420.45 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 153.54 | 12369.44 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 144.22 | 11739.74 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 139.90 | 11799.41 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 134.12 | 11914.27 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | Llama 3.1 8B | Q4_K_M | llama.cpp (CUDA) | 125.44 | 11309.70 | – | – | tg128 | 2026-04-04 | community | – | source |
| RTX 5090 | llama 7B | Q4_0 | llama.cpp (Vulkan) | 273.68 | 11796.38 | – | – | tg128 | 2025-10-11 | community | – | source |
| RTX 5090 | llama 7B | Q4_0 | llama.cpp (Vulkan) | 263.63 | 10381.64 | – | – | tg128 | 2025-10-11 | community | – | source |
| RTX 5090 | llama 7B | Q4_0 | llama.cpp (Vulkan) | 154.31 | 6011.18 | 160.0 | – | tg128 | 2026-09-17 | community | – | source |
| RTX 5090 | llama 7B | Q4_0 | llama.cpp (Vulkan) | 149.41 | 5546.39 | – | – | tg128 | 2026-09-17 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 58.61 | 3649 | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 56.73 | 3548 | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 55.84 | 3547 | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 55.60 | 3512 | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 55.16 | 3577 | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 51.69 | – | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 46.62 | – | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen 3.5 27B | Q6_K | llama.cpp (CUDA) | 38.88 | – | – | – | 32768 | 2026-04-01 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | BF16 | vLLM | 1404.7 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 1045.8 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 705.9 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 700.9 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 101.4 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 100.1 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | F16 | llama.cpp (CUDA) | 99.8 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen2.5-7B-Instruct 7B | BF16 | vLLM | 92.9 | – | – | – | – | 2026-08 | community | – | source |
| RTX 5090 | Qwen3 14B 14B | Q4_K | llama.cpp | 123.8 | 6497.6 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Qwen3 14B 14B | Q4_K | llama.cpp | 102.7 | 4473.1 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Qwen3 14B 14B | Q4_K | llama.cpp | 82.4 | 2908.4 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Qwen3 14B 14B | Q4_K | llama.cpp | 57.6 | 1707.2 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | Qwen3 14B 14B | Q4_K | llama.cpp | 37.2 | 908.4 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | Qwen3 32B 32B | Q4_K | llama.cpp | 61.4 | 2931.3 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Qwen3 32B 32B | Q4_K | llama.cpp | 50.9 | 2077.2 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Qwen3 32B 32B | Q4_K | llama.cpp | 43.8 | 1451.1 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 235.00 | – | – | – | tg128 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 234.35 | – | – | – | tg128 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 234.31 | – | – | – | tg128 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 215.63 | – | – | – | tg128 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 212.66 | – | – | – | 4096 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 211.93 | – | – | – | 4096 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 211.85 | – | – | – | 4096 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 171.85 | – | – | – | 16384 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 169.06 | – | – | – | 16384 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 168.94 | – | – | – | 16384 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 164.41 | – | – | – | 4096 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 134.39 | – | – | – | 32768 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 131.24 | – | – | – | 32768 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 131.09 | – | – | – | 32768 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 100.54 | – | – | – | 16384 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3 8B | Q4_K_M | llama.cpp (CUDA) | 65.07 | – | – | – | 32768 | 2026-10 | community | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 165.2 | 6605.2 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 148.3 | 6142.3 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 143.2 | 5611.4 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 133.5 | 4624.5 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 118.2 | 3242.6 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | Qwen3.5 35B 35B | MXFP4 | llama.cpp | 97.3 | 2003.7 | – | – | 262144 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K | llama.cpp | 58.8 | 3004.1 | – | – | 4096 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K | llama.cpp | 55.8 | 2721.5 | – | – | 16384 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K | llama.cpp | 53.8 | 2341.8 | – | – | 32768 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K | llama.cpp | 50.1 | 1606.4 | – | – | 65536 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K | llama.cpp | 44.1 | 1019.8 | – | – | 131072 | – | sourced | – | source |
| RTX 5090 | Qwen3.5-27B 27B | Q4_K_M | llama.cpp | 30.09 | 1515.83 | – | – | 1024 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.5-35B-A3B | Q4_K_S | llama.cpp | – | 12477.32 | – | – | 2048 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.5-35B-A3B | Q4_K_S | llama.cpp | – | 14524.81 | – | – | 2048 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.5-35B-A3B | Q4_K_S | llama.cpp | – | 12978.57 | – | – | 4096 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.5-35B-A3B | Q4_K_S | llama.cpp | – | 15298.0 | – | – | 4096 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.6-27B | Q4_K_M | llama.cpp | 136.59 | – | – | – | – | 2026-10-12 | sourced | – | source |
| RTX 5090 | Qwen3.6-27B | Q4_K_M | llama.cpp | 130.11 | – | – | – | – | 2026-10-12 | sourced | – | source |
| RTX 5090 | Qwen3.6-27B 27B hybrid GDN + full attention (MTP head) | NVFP4 | vLLM | 95.6 | – | – | – | – | 2026-07 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 344.1 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 333.9 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 329.4 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 307.6 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 301.7 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 292.8 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 281 | 8316 | – | – | tg128 | 2026-10-02 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_blend4 | llama.cpp (CUDA) | 258 | 8137 | – | – | tg128 | 2026-10-02 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_blend5 | llama.cpp (CUDA) | 258 | 8123 | – | – | tg128 | 2026-10-02 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | NVFP4 | llama.cpp (CUDA) | 250 | 8855 | – | – | tg128 | 2026-10-02 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_blend2 | llama.cpp (CUDA) | 247 | 8566 | – | – | tg128 | 2026-10-02 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 227.8 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.6-35B-A3B | Q4_K_M | llama.cpp (CUDA) | 187.0 | – | – | – | – | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 299.21 | – | – | – | 131072 | 2026-10-01 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 295.85 | – | – | – | 131072 | 2026-10-01 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 291.21 | – | – | – | 131072 | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 274.54 | – | – | – | 131072 | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B 27B | NVFP4 | llama.cpp (CUDA) | 152 | – | 575 | – | – | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q6_K | llama.cpp (focus-llama) | 102.7 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B 27B | NVFP4 | llama.cpp (CUDA) | 87.45 | 6258.74 | 575 | – | 128 | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | 82.8 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | 78.9 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 78.67 | – | – | – | – | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | 78.49 | – | – | – | – | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q6_K | llama.cpp | 76.2 | – | – | – | – | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 6122 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 8860 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | NInfer | – | 8466 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 8499 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 6518 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 7065 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | NVFP4 | llama.cpp (CUDA) | – | 7550 | – | – | 16384 | 2026-09 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 3731.18 | – | – | – | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 3718.85 | – | – | – | 2026-09-30 | community | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 4095.61 | – | – | 2048 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 4320.75 | – | – | 2048 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 4114.11 | – | – | 4096 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.8-27B | Q4_K_M | llama.cpp | – | 4353.09 | – | – | 4096 | 2026-10-13 | sourced | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | Strata | 221 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | llama.cpp (CUDA) | 218 | – | – | – | tg128 | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | llama.cpp (CUDA) | 210 | – | – | – | tg128 | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | Strata | 209 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | llama.cpp (CUDA) | 180 | – | – | – | tg128 | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | Strata | 140 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-M | Strata | 114 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | llama.cpp (CUDA) | 113 | 3000 | – | – | tg128 | 2026-10-04 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next | Q4_0 | llama.cpp (CUDA) | 67.8 | – | – | – | tg512 | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next | Q4_0 | llama.cpp (CUDA) | 54.5 | – | – | – | tg512 | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next | Q4_0 | llama.cpp (CUDA) | 30.8 | – | – | – | tg512 | 2026-10-03 | community | – | source |
| RTX 5090 | Qwen3.8-Flash-Next 126B | Gyro-S | Strata | – | 4883 | – | – | – | 2026-10-05 | community | – | source |
| RTX 5090 | qwen3moe 30B.A3B 30B-A3B | Q4_K_M | llama.cpp | 134.18 | 4081.2 | – | – | 1024 | 2026-09 | community | – | source |
| RTX 5090 | Ternary-Bonsai-2-27B | Ternary PQ2_0 | llama.cpp (CUDA) | 129.9 | 3893 | – | – | tg128 | 2026-09-16 | community | – | source |
| RTX 5090 | Ternary-Bonsai-2-27B | Ternary PTQ1_0 | llama.cpp (CUDA) | 120.5 | 1805 | – | – | tg128 | 2026-09-16 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085