M3 (GPU count not published, 16 GB)
2 records across 2 models (quants: Q4_K - Medium).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M3 (GPU count not published, 16 GB) | llama-3-1-8b-instruct | Q4_K - Medium | llamafile | 11.83115477777778 | 33.58097844444444 | – | 40.49901329166667 | – | – | sourced | – | source |
| M3 (GPU count not published, 16 GB) | llama-3-2-1b-instruct | Q4_K - Medium | llamafile | 61.584239 | 229.7350115555556 | – | 6.018140875 | – | – | sourced | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085