M4 (16 GB)
8 records across 4 models (quants: Q4_0, Q4_1, Q4_K - Medium, Q4_K - Small, Q6_K).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M4 (16 GB) | Devstral Small 2 24B | Q4_0 | llama.cpp | 3.3641 | – | – | 2.6874 | – | – | sourced | – | source |
| M4 (16 GB) | Devstral Small 2 24B | Q4_1 | llama.cpp | 0.0542 | – | – | 17.48 | – | – | sourced | – | source |
| M4 (16 GB) | Devstral Small 2 24B | Q4_K - Medium | llama.cpp | 0.0198 | – | – | 65.32 | – | – | sourced | – | source |
| M4 (16 GB) | Qwen3.5-27B | Q4_K - Medium | llama.cpp | 0.0084 | – | – | 66.9331 | – | – | sourced | – | source |
| M4 (16 GB) | Qwen3.5-35B-A3B | Q4_K - Medium | llama.cpp | 1.2956 | – | – | 13.0595 | – | – | sourced | – | source |
| M4 (16 GB) | Qwen3.5-9B | Q4_0 | llama.cpp | 4.0648 | – | – | 0.5057 | – | – | sourced | – | source |
| M4 (16 GB) | Qwen3.5-9B | Q4_K - Small | llama.cpp | 3.142 | – | – | 0.8445 | – | – | sourced | – | source |
| M4 (16 GB) | Qwen3.5-9B | Q6_K | llama.cpp | 2.1849 | – | – | 1.5031 | – | – | sourced | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085