Model family
MiniMax Models
MiniMax models are evaluated for long-context reasoning, coding, tool use, agents, and productivity workflows.
Best for
Reasoning
Use this family hub to compare MiniMax variants for reasoning workflows, then open the detail page for deeper deployment notes.
Agents
Use this family hub to compare MiniMax variants for agents workflows, then open the detail page for deeper deployment notes.
Code
Use this family hub to compare MiniMax variants for code workflows, then open the detail page for deeper deployment notes.
Source box
MiniMax family pages should verify exact M2/M2.7 checkpoint license, context, and availability from the official MiniMax model page before open-weight claims.
Verified through: June 2026
Model licenses, context windows, release names, and provider terms can vary by checkpoint. Verify the exact model card before production or commercial use.
Variants
MiniMax models grouped by workflow
Latest / flagship
MiniMax M2.7
MiniMax · MiniMax
Best for: Builders comparing current agent-ready open-weight models for coding and productivity workflows, for non-commercial or internally-authorized use.
MiniMax M3
MiniMax · MiniMax
Best for: Teams evaluating frontier-scale open-weight reasoning/coding models with genuine multimodal (image) input and a ~1M-token context window, willing to work within a named commercial license rather than a standard OSI one.
MiniMax M2.5
MiniMax · MiniMax
Best for: Builders tracking the M2 line's iteration for coding and agent workflows; M2.7 supersedes it with a larger context window if a newer release is available.
MiniMax M2.1
MiniMax · MiniMax
Best for: Builders tracking the M2 line's iteration for coding and agent workflows; M2.5 or M2.7 supersede it if a newer release is available.
Coding
Reasoning
Compare
All MiniMax models in the directory
| Model | Type | Best for | Local runner notes | License | Detail |
|---|---|---|---|---|---|
| MiniMax M2.7 | Reasoning | Builders comparing current agent-ready open-weight models for coding and productivity workflows, for non-commercial or internally-authorized use. | 228.7B total parameters, 8 experts active per token. No GGUF in the official repository -- server-class multi-GPU or hosted inference only. | MiniMax M2 Non-Commercial License (modified MIT base) | Open |
| MiniMax M3 | Reasoning | Teams evaluating frontier-scale open-weight reasoning/coding models with genuine multimodal (image) input and a ~1M-token context window, willing to work within a named commercial license rather than a standard OSI one. | 427B total parameters, sparse MoE. No GGUF exists in the official repository and the custom MiniMaxM3Sparse architecture class has no known llama.cpp support, so local consumer-hardware deployment is not currently possible regardless of quantization. Multi-GPU server inference (vLLM, SGLang) or a hosted provider only. | MiniMax Community License | Open |
| MiniMax M2.5 | Reasoning | Builders tracking the M2 line's iteration for coding and agent workflows; M2.7 supersedes it with a larger context window if a newer release is available. | 228.7B total parameters, 8 experts active per token. No GGUF in the official repository -- server-class multi-GPU or hosted inference only. | MiniMax M2 Non-Commercial License (modified MIT base) | Open |
| MiniMax M2.1 | Reasoning | Builders tracking the M2 line's iteration for coding and agent workflows; M2.5 or M2.7 supersede it if a newer release is available. | 228.7B total parameters, 8 experts active per token. No GGUF in the official repository -- server-class multi-GPU or hosted inference only. | MiniMax M2 Non-Commercial License (modified MIT base) | Open |
| MiniMax M2 | Reasoning | Builders who want the baseline M2-generation release for comparison against later M2.x point releases or M3. | 228.7B total parameters, 8 experts active per token. No GGUF in the official repository -- server-class multi-GPU or hosted inference only. | MiniMax M2 Non-Commercial License (modified MIT base) | Open |
| MiniMax M1 | Reasoning | Builders who specifically want a permissively-licensed (Apache 2.0) MiniMax release for commercial use, at the cost of a larger, older architecture than the M2.x/M3 lines. | 456B total parameters, 2 experts active per token. No GGUF in the official repository -- server-class multi-GPU or hosted inference only. | Apache 2.0 | Open |