MiniMax-M3
MiniMax-M3 is an open-weights language model from MiniMax, released June 1, 2026. It has 427B parameters (mixture-of-experts, 5.8B active). The context window is 1M tokens. On the Epoch Capabilities Index it scores 147.7, rank 79 of 256 rated models. API price: $0.3 input / $1.2 output per million tokens.
MiniMax M3 is officially released today. M3 reaches frontier-level performance on specialized tasks such as coding and agentic work. It uses MSA (MiniMax Sparse Attention), a new attention architecture proposed by our team, and supports ultra-long context windows of up to 1M tokens.
Check it against your own PC →
Can you run MiniMax-M3 locally?
Even with 32 GB of VRAM and 32 GB of RAM it is too large to run at a usable speed on a single consumer GPU. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Too large | – | – | – | – |
| 8 GB VRAM | Too large | – | – | – | – |
| 12 GB VRAM | Too large | – | – | – | – |
| 16 GB VRAM | Too large | – | – | – | – |
| 24 GB VRAM | Too large | – | – | – | – |
| 32 GB VRAM | Too large | – | – | – | – |
Ready-made quantizations on Hugging Face: unsloth/MiniMax-M3-GGUF · bartowski/MiniMax-M3-GGUF · DevQuasar/MiniMaxAI.MiniMax-M3-GGUF · AesSedai/MiniMax-M3-GGUF · vultr/MiniMax-M3-GGUF · kaitchup/MiniMax-M3-GGUF-MoQ
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Reasoning | 11.3 | Chess Puzzles, SimpleBench, Mystery Game Puzzles |
| Math | 34.4 | ProofBench, OTIS Mock AIME 2024-2025 |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| GPQA Diamond | 91.5% |
| OTIS Mock AIME | 71.1% |
| SimpleBench | 35.0% |
| ProofBench | 18.0% |
| Chess Puzzles | 12.7% |
| OSWorld 2.0 | 4.6% |
| Mystery Game Puzzles | 0.0% |
Links
Announcement · Hugging Face: MiniMaxAI/MiniMax-M3 · Epoch AI benchmarks
Models with a similar score
Kimi K2.6 · GPT-5 Pro · Kimi K2.7 Code · GPT-5.1 · Beam 1.0 · Inkling-Small · GLM-5.1 · DeepSeek-V4-Pro
More from MiniMax
MiniMax-M3.1-Flash · MiniMax-M2.7 · MiniMax-M2.5 · MiniMax-M2.1 · MiniMax-M2 · MiniMax-M1-80k · MiniMax-M1-40k