Can I run MiniMax M3 locally?
MiniMax M3 by MiniMaxAI has 427B parameters (20B active per token, 128 experts, 4 used per token). At 4-bit with an 8K context it needs about 248 GB of GPU memory. It fits on 2 of 65 GPUs and Macs we track.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from unsloth/MiniMax-M3-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx | 128K ctx |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | rough | 150 GB | 151 GB | 154 GB | 165 GB |
| 3-bit (Q3_K_M) • | usable | 183 GB | 184 GB | 187 GB | 198 GB |
| 4-bit (Q4_K_M) • | good, most popular | 248 GB | 248 GB | 251 GB | 263 GB |
| 5-bit (Q5_K_M) • | very good | 298 GB | 299 GB | 301 GB | 313 GB |
| 6-bit (Q6_K) • | near lossless | 332 GB | 333 GB | 335 GB | 347 GB |
| 8-bit (Q8_0) • | effectively lossless | 424 GB | 424 GB | 427 GB | 438 GB |
| 16-bit (F16/BF16) • | original weights | 796 GB | 796 GB | 799 GB | 810 GB |
Which GPUs and Macs run MiniMax M3?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
Prefer an API?
MiniMax: MiniMax M3 is on OpenRouter at $0.30 per 1M input tokens and $1.20 per 1M output tokens. Price details and history →
Questions
How much VRAM does MiniMax M3 need?
About 248 GB at 4-bit (Q4_K_M) with an 8K context: 247 GB of weights (real GGUF file size), 0.9 GB of KV cache and 0.8 GB of runtime overhead. At 8-bit it needs about 424 GB.
What is the smallest GPU that runs MiniMax M3?
No single consumer GPU in our table fits it at Q4_K_M with 8K context; the smallest option that does is the Mac M3 Ultra 512GB (512 GB).
Can I run MiniMax M3 on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M3 Ultra 512GB or larger, using the default macOS GPU memory limit.
Model facts
| Hugging Face repo | MiniMaxAI/MiniMax-M3 |
| Parameters | 427,040,140,160 (427B) · ~20B active (computed from config) |
| Layers / hidden size | 60 / 6144 |
| Attention | GQA; 64 heads, 4 KV heads, head dim 128 |
| KV cache per token (FP16) | 120.0 KiB |
| Max context | 1,048,576 tokens |
| License | other |
| Released on HF | Jun 2, 2026 |
| Downloads (30d) / likes | 126,961 / 1,605 |
Sources: Hugging Face model API · config.json (MiniMaxAI/MiniMax-M3) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.