LLM
GA
Open weights
MiniMax M2.5
Cheap, fast agentic coding
Release2026-01
Context205K
Max output—
Parameters230B
Local / VRAM240–320 GB
LicenseOpen weights
Overview
MiniMax M2.5 is a model in the LLM category from MiniMax. It was released in 2026-01. The vendor is based in China. Best for: Cheap, fast agentic coding.
Strengths
- 10B active
- Strong coding
- Open weights
Limits
- No vision
- 200K context
Specifications
| Vendor | MiniMax |
|---|---|
| Category | LLM |
| Release | 2026-01 |
| Status | GA |
| License | Open weights |
| License name | Modified MIT |
| Commercial use | Allowed |
| Parameters | 230B |
| Active parameters | 10B |
| Architecture | MoE transformer |
| Context window | 205K |
| Input modalities | Text |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| VRAM | 240–320 GB |
| System RAM | 256–512 GB |
| API access | Yes |
| Pricing | Free weights / API |
| Confidence | High |
| Vendor country | China |
| Vendor type | Open lab |
Pricing & hardware
Free weights / API
This model can be self-hosted or run locally. Approximate GPU memory: 240–320 GB. Approximate system RAM: 256–512 GB. API access is available.