2x RTX 3090
54 records across 5 models (quants: 3.0-bit, IQ3_XXS, IQ4_XS, IQ4_XS-PLEQ4, Q4_K_M, Q5_K_M, UD-IQ4_XS, UD-Q6_K_XL, UD-Q8_K_XL, not stated).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 13.57 | – | – | – | 65535 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 13.20 | – | – | – | 262144 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 13.12 | – | – | – | 262144 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 13.03 | – | – | – | 405000 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 12.7 | – | – | – | 65535 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 12.57 | – | – | – | 405000 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 10.96 | – | – | – | 1048576 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 9.71 | – | – | – | 1048576 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 7.53 | – | – | – | 1048576 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 7.08 | – | – | – | 405000 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | DeepSeek-V4-Flash 284B-A13B | UD-Q8_K_XL | llama.cpp (CUDA) | 6.88 | – | – | – | 65535 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 22.4 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | GLM-5.3-Flash | 3.0-bit | llama.cpp (CUDA) | 11.9 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | GLM-5.3-Flash | Q4_K_M | llama.cpp (CUDA) | 7.90 | – | – | – | 262144 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | GLM-5.3-Flash | Q4_K_M | llama.cpp (CUDA) | 7.72 | – | – | – | 262144 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | MiMo-V2.6-Flash | IQ3_XXS | llama.cpp (CUDA) | 10.9 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | MiMo-V2.6-Flash | IQ3_XXS | llama.cpp (CUDA) | 4.6 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | Qwen3.8-27B | not stated | llama.cpp (CUDA) | 59.98 | 1848.8 | – | – | 262144 | 2026-09-27 | community | – | source |
| 2x RTX 3090 | Qwen3.8-27B | not stated | llama.cpp (CUDA) | 57.98 | 1132.0 | – | – | 262144 | 2026-09-27 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 125B | UD-IQ4_XS | llama.cpp (CUDA) | 59.6 | – | – | – | 4096 | 2026-10-04 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 49.3 | – | – | – | tg3500 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 125B | UD-IQ4_XS | llama.cpp (CUDA) | 47.6 | – | – | – | 4096 | 2026-10-04 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next | UD-IQ4_XS | llama.cpp (CUDA) | 46.5 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 42 | – | – | – | tg3000 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 41.2 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 40.7 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS-PLEQ4 | llama.cpp (CUDA) | 38.8 | – | – | – | 1048576 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS-PLEQ4 | llama.cpp (CUDA) | 37.7 | – | – | – | 1048576 | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 36.6 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 35.5 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 33.3 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 30.4 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 28.6 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 27.9 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next | UD-IQ4_XS | llama.cpp (CUDA) | 27.7 | – | – | – | – | 2026-09-25 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 27.3 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 26.8 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 26.8 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 26.2 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 25.5 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 24.4 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 24.3 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 23.6 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 23.6 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 23.0 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 21.5 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q4_K_M | llama.cpp (CUDA) | 20.0 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 16.8 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 16.2 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 15.5 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | Q5_K_M | llama.cpp (CUDA) | 14.2 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 13.9 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | UD-Q6_K_XL | llama.cpp (CUDA) | 12.8 | – | – | – | – | 2026-10-05 | community | – | source |
| 2x RTX 3090 | Qwen3.8-Flash-Next 177B | IQ4_XS | llama.cpp (CUDA) | 11.9 | – | – | – | – | 2026-10-05 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085