P106-100 × Llama 3.2 1B Instruct
Llama 3.2 1B Instruct on P106-100
78.8best gen tok/s
1,879best prompt tok/s
4,352max context tested
7results
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 | 78.8 | 1,874 | LocalScore ↗ | 512+128 | 720 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.7 | 1,874 | LocalScore ↗ | 4,352 | 724 ms | llamafile 0.9.2 |
| Q4_K - Medium | 78.7 | 1,879 | LocalScore ↗ | 512+128 | 721 ms | llamafile 0.9.2 |
| Q4_K - Medium | 73.9 | 1,812 | LocalScore ↗ | 512+128 | 739 ms | llamafile 0.9.2 |
| Q4_K - Medium | 73.9 | 1,823 | LocalScore ↗ | 512+128 | 730 ms | llamafile 0.9.2 |
| Q4_K - Medium | 54.8 | 1,525 | LocalScore ↗ | 512+128 | 848 ms | llamafile 0.9.2 |
| Q4_K - Medium | 53.3 | 1,520 | LocalScore ↗ | 512+128 | 855 ms | llamafile 0.9.2 |
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.