M3 (10-core GPU, 24 GB)
3 records across 3 models (quants: Q4_K - Medium).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M3 (10-core GPU, 24 GB) | llama-3-1-8b-instruct | Q4_K - Medium | llamafile | 10.22738411111111 | 109.8980422777778 | – | 11.670486025444445 | – | – | sourced | – | source |
| M3 (10-core GPU, 24 GB) | llama-3-2-1b-instruct | Q4_K - Medium | llamafile | 64.71307648888889 | 916.5074562888888 | – | 1.4258484739777777 | – | – | sourced | – | source |
| M3 (10-core GPU, 24 GB) | qwen-2-5-14b-instruct | Q4_K - Medium | llamafile | 6.133619851851852 | 65.80348074074074 | – | 19.91686526244444 | – | – | sourced | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085