LLM
GA
Open weights
MiniMax M2.7
Research and non-commercial agentic coding
Release2026-03-18
Context205K
Max output—
Parameters229B
Local / VRAMLocal
LicenseOpen weights
Overview
MiniMax M2.7 is a model in the LLM category from MiniMax. It was released in 2026-03-18. The vendor is based in China. Best for: Research and non-commercial agentic coding.
Notes: Superseded by MiniMax M3.
Hardware: Weights on Hugging Face; VRAM figures not verified.
Strengths
- Strong coding
- Open weights
Limits
- Non-commercial license for the weights
- Text only
- 200K context
Specifications
| Vendor | MiniMax |
|---|---|
| Category | LLM |
| Release | 2026-03-18 |
| Status | GA |
| License | Open weights |
| License name | MiniMax M2.7 Non-Commercial License |
| Commercial use | Not allowed |
| Parameters | 229B |
| Architecture | MoE transformer |
| Context window | 205K |
| Input modalities | Text |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| API access | Yes |
| Pricing | $0.30 / $1.20 per 1M token |
| Confidence | Medium |
| Vendor country | China |
| Vendor type | Open lab |
Pricing & hardware
$0.30 / $1.20 per 1M token
This model can be self-hosted or run locally. API access is available.