MiniMax-M2.7
By MiniMax · China
Updated 2026-07-13
Overview
MiniMax's agentic MoE with 229B total and 10B active parameters under Apache 2.0. Open-sourced April 12, 2026 and currently the top trending model on Hugging Face.
When to pick this model
- Multi-step agentic workflows and tool use
- Terminal and shell automation pipelines
- SWE-Bench-style autonomous coding agents
- Long-context tasks needing 200K+ tokens
- Commercial agent products under Apache 2.0
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 138 GB |
| Q5_K_M | 165 GB |
| Q8_0 | 246 GB |
| FP16 (no quantization) | 458 GB |
VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.
In practice, MiniMax-M2.7 is server-class even at Q4_K_M (138 GB). Stepping up to Q8_0 nearly doubles the footprint to 246 GB, and unquantized FP16 weights take 458 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, MiniMax-M2.7 needs roughly 160 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 2 tokens/sec on entry-level GPUs, on the order of 10 tokens/sec on a mid-range card, and up to 25 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches MiniMax-M2.7 to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.
| GPU memory | Example cards | Best fit for MiniMax-M2.7 |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | Does not fit — needs 138 GB at Q4_K_M |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | Does not fit — needs 138 GB at Q4_K_M |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | Does not fit — needs 138 GB at Q4_K_M |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | Does not fit — needs 138 GB at Q4_K_M |
| 32 GB | RTX 5090 | Does not fit — needs 138 GB at Q4_K_M |
Which GPU should you buy to run MiniMax-M2.7?
To run MiniMax-M2.7 locally at Q4, you need ~138 GB of VRAM. The best value for this is a Apple Mac Studio (64+ GB unified memory).
As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.
Published benchmark scores
| Benchmark | Score |
|---|---|
| SWE-Pro | 56.22 |
| Terminal Bench | 57 |
Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.
Strengths
- State-of-the-art open agentic performance
- 56.22% SWE-Bench Pro and 57% Terminal Bench
- Only 10B active parameters keeps inference fast
- Clean Apache 2.0 license
- #1 trending on Hugging Face at launch
Limitations
- 138GB+ in Q4 demands serious server hardware
- Verbose output in agent mode inflates token costs
- Newer release means thinner tooling ecosystem
Typical workloads
In our catalog grid, MiniMax-M2.7 is filed under Autonomous Agents, Reasoning, Code — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: multi-step reasoning and math-flavoured tasks.
The 200k-token context window is large enough to hold entire codebases' worth of files or long reports in a single prompt, which is what makes local RAG and document analysis practical. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.
Architecture & training
Architecture: MoE 229B/10B active · 205k ctx · self-evolving agent
Training: Open-sourced April 12, 2026. Successor to M2.5.
The most exciting agentic open-weight model of 2026, if your hardware can host 229B parameters.
Quick start
# HuggingFace (GGUF) : unsloth/MiniMax-M2.7-GGUFOr use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.
Similar models worth comparing
Frequently asked questions
How much VRAM does MiniMax-M2.7 need?
At the recommended Q4_K_M quantization, MiniMax-M2.7 needs about 138 GB of VRAM. Q8_0 takes 246 GB, and unquantized FP16 weights take 458 GB.
Can MiniMax-M2.7 run without a GPU?
Yes — with roughly 160 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.
What context window does MiniMax-M2.7 support?
MiniMax-M2.7 supports a 200k-token context window (205,000 tokens).
Can I use MiniMax-M2.7 commercially?
Yes. MiniMax-M2.7 is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is MiniMax-M2.7 on consumer hardware?
Our compatibility engine estimates on the order of 10 tokens/sec on a mid-range GPU and up to 25 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of MiniMax-M2.7 should I download first?
Start with Q4_K_M (138 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It does not fit a single 24 GB consumer card — plan for multi-GPU or server hardware.