RTX 3080 × Meta Llama 3.1 8B Instruct
Meta Llama 3.1 8B Instruct on RTX 3080
93.3best gen tok/s
3,156best prompt tok/s
4,352max context tested
14results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 93.3 | 2,950 | LocalScore ↗ | 4,352 | 431 ms | llamafile 0.9.2 |
| Q4_K - Medium | 92.5 | 3,140 | LocalScore ↗ | 512+128 | 409 ms | llamafile 0.9.2 |
| Q4_K - Medium | 92.4 | 2,796 | LocalScore ↗ | 4,352 | 453 ms | llamafile 0.9.2 |
| Q4_K - Medium | 92.1 | 3,156 | LocalScore ↗ | 512+128 | 406 ms | llamafile 0.9.2 |
| Q4_K - Medium | 89.7 | 2,880 | LocalScore ↗ | 512+128 | 432 ms | llamafile 0.9.2 |
| Q4_K - Medium | 80.3 | 3,123 | LocalScore ↗ | 512+128 | 411 ms | llamafile 0.9.3 |
| Q4_K - Medium | 79.0 | 3,125 | LocalScore ↗ | 512+128 | 411 ms | llamafile 0.9.3 |
| Q4_K - Medium | 76.6 | 3,028 | LocalScore ↗ | 512+128 | 421 ms | llamafile 0.9.2 |
| Q4_K - Medium | 76.4 | 3,128 | LocalScore ↗ | 4,352 | 410 ms | llamafile 0.9.3 |
| Q4_K - Medium | 75.7 | 2,792 | LocalScore ↗ | 512+128 | 441 ms | llamafile 0.9.2 |
| Q4_K - Medium | 75.5 | 2,891 | LocalScore ↗ | 4,352 | 444 ms | llamafile 0.9.3 |
| Q4_K - Medium | 74.8 | 3,079 | LocalScore ↗ | 512+128 | 415 ms | llamafile 0.9.3 |
| Q4_K - Medium | 68.2 | 2,836 | LocalScore ↗ | 512+128 | 452 ms | llamafile 0.9.2 |
| Q4_K - Medium | 52.1 | 2,963 | LocalScore ↗ | 4,352 | 430 ms | llamafile 0.9.3 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
Includes results from LocalScore, a Mozilla Builders project by cjpais, mirrored in the cn0303/local-llm-speed-benchmarks dataset. LocalScore runs use llamafile, which is built on llama.cpp.