36GB VRAM
Mac M3 Pro 36G — what LLMs can it run?
64 of 79 indexed models fit comfortably in 36GB at 4K context, each at the highest-quality quant that still leaves headroom.
32B · 18
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| Mixtral 8x7B Instruct47B MoE | Q4_K_M | 30.13 GB | +5.9 GB | 48 |
| Seed-OSS 36B Instruct36B | Q5_K_M | 27.81 GB | +8.2 GB | 37 |
| Command R 35B35B | Q4_K_M | 22.86 GB | +13.1 GB | 42 |
| Yi 1.5 34B Chat34B | Q4_K_M | 22.82 GB | +13.2 GB | 40 |
| Qwen3 32B Instruct32B | Q4_K_M | 21.85 GB | +14.2 GB | 42 |
| Qwen2.5 32B Instruct32B | Q4_K_M | 21.69 GB | +14.3 GB | 44 |
| Qwen2.5-Coder 32B Instruct32B | Q4_K_M | 21.69 GB | +14.3 GB | 44 |
| DeepSeek-R1-Distill-Qwen-32B32B | Q4_K_M | 21.69 GB | +14.3 GB | 42 |
| Qwen3 30B-A3B Instruct30B-A3B | Q5_K_M | 23.04 GB | +13 GB | 82 |
| Qwen3-Coder 30B-A3B Instruct30B-A3B | Q5_K_M | 23.04 GB | +13 GB | 80 |
| Qwen3-VL 30B-A3B Instruct30B-A3B | Q4_K_M | 19.73 GB | +16.3 GB | 95 |
| Gemma 3 27B IT27B | Q4_K_M | 19.49 GB | +16.5 GB | 48 |
| Gemma 2 27B Instruct27B | Q5_K_M | 21.76 GB | +14.2 GB | 42 |
| Mistral Small 24B Instruct24B | EXL2 4.65bpw | 15.26 GB | +20.7 GB | 88 |
| Devstral Small 1.1 24B24B | Q6_K | 20.91 GB | +15.1 GB | 48 |
| Magistral Small 1.2 24B24B | Q6_K | 20.91 GB | +15.1 GB | 47 |
| Codestral 22B22B | Q4_K_M | 15.03 GB | +21 GB | 58 |
| GPT-OSS 20B21B MoE | MXFP4 | 11.81 GB | +24.2 GB | 195 |
14B · 17
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| InternLM2 20B Chat20B | Q5_K_M | 15.59 GB | +20.4 GB | 68 |
| DeepSeek-Coder-V2-Lite Instruct16B | Q8_0 | 18.36 GB | +17.6 GB | 118 |
| DeepSeek-V2-Lite Chat16B | Q4_K_M | 10.87 GB | +25.1 GB | 142 |
| StarCoder2 15B15B | Q4_K_M | 10.19 GB | +25.8 GB | 92 |
| Qwen3 14B Instruct14B | Q5_K_M | 11.65 GB | +24.4 GB | 78 |
| Qwen2.5 14B Instruct14B | Q5_K_M | 11.73 GB | +24.3 GB | 86 |
| DeepSeek-R1-Distill-Qwen-14B14B | EXL2 4.65bpw | 9.75 GB | +26.3 GB | 128 |
| Phi-4 14B14B | Q5_K_M | 11.77 GB | +24.2 GB | 78 |
| Phi-3 Medium 14B Instruct14B | Q6_K | 12.86 GB | +23.1 GB | 88 |
| Mistral Nemo 12B Instruct12B | Q6_K | 11.14 GB | +24.9 GB | 95 |
| Gemma 3 12B IT12B | Q5_K_M | 10.6 GB | +25.4 GB | 92 |
| Stable LM 2 12B Chat12B | Q4_K_M | 8.35 GB | +27.7 GB | 108 |
| Jamba 1.5 Mini12B | Q4_K_M | 8.15 GB | +27.9 GB | 95 |
| Llama 3.2 11B Vision Instruct11B | Q8_0 | 12.9 GB | +23.1 GB | 72 |
| Solar 10.7B Instruct11B | Q4_K_M | 7.6 GB | +28.4 GB | 125 |
| Falcon 3 10B Instruct10B | Q4_K_M | 7.28 GB | +28.7 GB | 118 |
| Gemma 2 9B Instruct9B | Q8_0 | 11.7 GB | +24.3 GB | 108 |
7B · 19
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| GLM-4-9B-Chat9B | Q8_0 | 10.16 GB | +25.8 GB | 105 |
| Qwen3-VL 8B Instruct8B | Q8_0 | 10.39 GB | +25.6 GB | 108 |
| Qwen2-VL 7B Instruct7B | Q4_K_M | 5.49 GB | +30.5 GB | 72 |
| Granite 3.1 8B Instruct8B | Q4_K_M | 5.88 GB | +30.1 GB | 142 |
| Qwen3 8B Instruct8B | Q6_K | 7.64 GB | +28.4 GB | 122 |
| Llama 3.1 8B Instruct8B | Q8_0 | 9.47 GB | +26.5 GB | 118 |
| Nous Hermes 3 Llama 3.1 8B8B | EXL2 4.65bpw | 5.43 GB | +30.6 GB | 232 |
| Aya 23 8B8B | Q4_K_M | 5.64 GB | +30.4 GB | 145 |
| OpenChat 3.6 8B8B | EXL2 4.65bpw | 5.43 GB | +30.6 GB | 228 |
| DeepSeek-R1-Distill-Llama-8B8B | Q5_K_M | 6.51 GB | +29.5 GB | 128 |
| InternLM2 7B Chat7B | Q4_K_M | 5.45 GB | +30.6 GB | 148 |
| Qwen2.5 7B Instruct7B | Q6_K | 6.77 GB | +29.2 GB | 132 |
| Qwen2.5-Coder 7B Instruct7B | EXL2 4.65bpw | 4.87 GB | +31.1 GB | 248 |
| WizardLM-2 7B7B | Q4_K_M | 5.07 GB | +30.9 GB | 152 |
| DeepSeek-R1-Distill-Qwen-7B7B | EXL2 4.65bpw | 4.87 GB | +31.1 GB | 210 |
| OLMo 2 7B Instruct7B | Q8_0 | 10.3 GB | +25.7 GB | 125 |
| Mistral 7B Instruct v0.37B | Q6_K | 6.75 GB | +29.3 GB | 135 |
| Zephyr 7B Beta7B | Q6_K | 6.75 GB | +29.3 GB | 132 |
| Gemma 3 4B IT4B | Q8_0 | 5.36 GB | +30.6 GB | 145 |
≤3B · 10
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| Qwen3 4B Instruct4B | Q6_K | 4.05 GB | +32 GB | 145 |
| Phi-4 Mini Instruct3.8B | Q8_0 | 4.68 GB | +31.3 GB | 262 |
| Phi-3.5 Mini Instruct3.8B | Q8_0 | 5.89 GB | +30.1 GB | 255 |
| Llama 3.2 3B Instruct3B | Q8_0 | 4.05 GB | +32 GB | 285 |
| Qwen2.5 3B Instruct3B | Q8_0 | 3.67 GB | +32.3 GB | 290 |
| Gemma 2 2B Instruct2B | Q8_0 | 3.34 GB | +32.7 GB | 320 |
| Qwen3 1.7B Instruct1.7B | Q8_0 | 2.39 GB | +33.6 GB | 240 |
| Qwen2.5 1.5B Instruct1.5B | Q8_0 | 1.83 GB | +34.2 GB | 410 |
| Llama 3.2 1B Instruct1B | Q8_0 | 1.51 GB | +34.5 GB | 450 |
| Qwen2.5 0.5B Instruct0.5B | Q8_0 | 0.6 GB | +35.4 GB | 540 |
How this list is built
Each row is the lowest-perplexity-loss quant of that model whose estimated total — weights plus KV cache at 4K context plus activation buffer — uses at most 88% of the card. That is the calculator's "green" threshold, so every row here has real headroom rather than only just fitting. Raise the context length and the list shortens; the calculator lets you check any combination directly.
Stepping up
A Mac M3 Max 48G (48GB) fits 5 more of the indexed models than this card. Mac M3 Max 48G →
64 of 79 indexed models fit comfortably in 36GB at 4K context, each at the highest-quality quant that still leaves headroom.