All hardware / M4 32GB
With 32 GB unified memory, the M4 32GB runs 14 of 22 popular open models fully in fast memory at Q4_K_M or better (8K context). The largest that fit are Qwen3.6 35B-A3B (Q4_K_M), Gemma 4 31B (Q5_K_M), Granite 4.2 30B (Q5_K_M). Fastest: Qwen3.6 35B-A3B at about 40 tokens/sec and Gemma 4 E2B at about 29 tokens/sec.
Check with your own RAM and context length| Model | Result | Best quant | Memory needed | Est. speed |
|---|---|---|---|---|
| Qwen3.6 35B-A3B | Runs on GPU | Q4_K_M | 22.2 GB | 40 tok/s |
| Gemma 4 E2B | Runs on GPU | Q8_0 | 6.1 GB | 29 tok/s |
| gpt-oss-20b | Runs on GPU | Q6_K | 18.1 GB | 24 tok/s |
| Gemma 4 26B A4B | Runs on GPU | Q6_K | 21.8 GB | 23 tok/s |
| Ministral 3 3B | Runs on GPU | Q8_0 | 4.8 GB | 18 tok/s |
| Gemma 4 E4B | Runs on GPU | Q8_0 | 9.3 GB | 15 tok/s |
| Qwen3.5 9B | Runs on GPU | Q8_0 | 11.3 GB | 7.0 tok/s |
| Gemma 4 12B | Runs on GPU | Q8_0 | 13.8 GB | 5.6 tok/s |
| Phi-4 14B | Runs on GPU | Q8_0 | 17.7 GB | 4.6 tok/s |
| Qwen3.6 27B | Runs on GPU | Q5_K_M | 20.3 GB | 3.7 tok/s |
| Mistral Small 3.1 24B | Runs on GPU | Q6_K | 21.5 GB | 3.7 tok/s |
| Qwen3.8 27B | Runs on GPU | Q5_K_M | 20.9 GB | 3.6 tok/s |
| Granite 4.2 30B | Runs on GPU | Q5_K_M | 21.6 GB | 3.5 tok/s |
| Gemma 4 31B | Runs on GPU | Q5_K_M | 23.0 GB | 3.3 tok/s |
| Hermes 4.3 36B | Lower quality fit | Q3_K_M | 19.7 GB | 4.1 tok/s |
| Llama 3.3 70B | Too big | Q4_K_M | 45.9 GB | — |
| Llama 4 Scout | Too big | Q4_K_M | 68.1 GB | — |
| gpt-oss-120b | Too big | Q4_K_M | 72.0 GB | — |
| Qwen3.5 122B-A10B | Too big | Q4_K_M | 75.3 GB | — |
| Qwen3.8-Flash-Next | Too big | Q4_K_M | 107.8 GB | — |
| GLM-5.3-Flash | Too big | Q4_K_M | 196.0 GB | — |
| DeepSeek V4.1-Flash | Too big | Q4_K_M | 335.8 GB | — |
OllamaLM StudioJanGPT4AllAnythingLLMllama.cppMLX LMKoboldCppText Generation WebUILocalAI