MiniMax model product
MiniMax M2
MiniMax M2 is an open-source large language model by MiniMax, built for agents and coding tasks.
Updated Aug 12, 2026. Default version: MiniMax M2
LLMBoard score43.9MiniMax M2
Coverage80%15 benchmark families
Context window196.6KTokens
Official input price$0.30MiniMax (minimax.io) API
MiniMax M2 Specifications
Technical details for the model's default version.
- Version
- MiniMax M2
- Released
- Oct 27, 2025
- Knowledge cutoff
- Unknown
- Parameters
- 230B
- Context window
- 196.6K
- Max output
- 128K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- MIT
MiniMax M2 Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
MiniMax M2 Benchmark Results
Benchmark scores for MiniMax M2.
| IF | 72.0% | 02 | 2 | 0.0% | C | |
| AA-Index | 61.0% | 03 | 3 | 0.0% | C | |
| Terminal-Bench | 46.3% | 04 | 25 | 87.5% | C | |
| LiveCodeBench | 83.0% | 05 | 73 | 94.4% | C | |
| Multi-SWE-Bench | 36.2% | 05 | 6 | 20.0% | C | |
| Tau-bench | 77.2% | 05 | 6 | 20.0% | C | |
| BrowseComp-zh | 48.5% | 11 | 13 | 16.7% | C | |
| Tau2 Telecom | 87.0% | 17 | 35 | 52.9% | C | |
| SciCode | 36.0% | 18 | 19 | 5.6% | C | |
| SWE-bench Multilingual | 56.5% | 27 | 34 | 21.2% | C | |
| MMLU-Pro | 82.0% | 39 | 129 | 70.3% | C | |
| BrowseComp | 44.0% | 50 | 58 | 14.0% | C | |
| SWE-Bench Verified | 69.4% | 63 | 105 | 40.4% | C | |
| AIME 2025 | 78.0% | 75 | 114 | 34.5% | C | |
| Humanity's Last Exam | 12.5% | 77 | 93 | 17.4% | C | |
| GPQA | 78.0% | 95 | 234 | 59.7% | C |
MiniMax M2 Arena Results
Preference and agent-evaluation results for the default version.
| webdev | overall | 91 | 1296.7 | 6,537 | N/A | |
| text factuality | overall | 150 | 1362.7 | 6,741 | N/A | |
| text | overall | 183 | 1342.4 | 6,909 | N/A | |
| text style control | overall | 201 | 1346.0 | 6,909 | N/A |
MiniMax M2 Runtime Performance
Provider-specific output speed and catalog latency for MiniMax M2. Runtime does not affect the capability score.
| MiniMax | 318.387 tok/s | 1.591 s | 1M | 1M |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
MiniMax M2 Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $0.30 input, $1.2 output per 1M
- Official provider
- MiniMax (minimax.io)
- Lowest third-party
- From $0.17 input, $1.53 output per 1M via NanoGPT
- Tracked offerings
- 17
| MiniMax Token Plan (minimax.io) | MiniMax-M2 | global | N/A | N/A | 196.6K | |
| MiniMax Token Plan (minimaxi.com) | MiniMax-M2 | global | N/A | N/A | 196.6K | |
| NanoGPT | MiniMax-M2 | global | $0.17 | $1.53 | 200K | |
| LLM Gateway | minimax-m2 | global | $0.20 | $1 | 196.6K | |
| OpenRouter | minimax/minimax-m2 | global | $0.255 | $1.02 | 204.8K | |
| ZenMux | minimax/minimax-m2 | global | $0.30 | $1.2 | 204K | |
| Amazon Bedrock | minimax.minimax-m2 | global | $0.30 | $1.2 | 204.6K | |
| MiniMax (minimax.io) | MiniMax-M2 | global | $0.30 | $1.2 | 196.6K | |
| Kilo Gateway | minimax/minimax-m2 | global | $0.30 | $1.2 | 204.8K | |
| Merge Gateway | minimax/minimax-m2 | global | $0.30 | $1.2 | 204.8K | |
| Vercel AI Gateway | minimax/minimax-m2 | global | $0.30 | $1.2 | 205K | |
| NovitaAI | minimax/minimax-m2 | global | $0.30 | $1.2 | 204.8K | |
| Hugging Face | MiniMaxAI/MiniMax-M2 | global | $0.30 | $1.2 | 204.8K | |
| MiniMax (minimaxi.com) | MiniMax-M2 | global | $0.30 | $1.2 | 196.6K | |
| 302.AI | MiniMax-M2 | global | $0.33 | $1.32 | 1M | |
| Cortecs | minimax-m2 | global | $0.349 | $1.41 | 400K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
MiniMax M2 Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| MiniMax M2 | 43.9 | 230B | 196.6K | 128K | Yes | MIT |
What is MiniMax M2?
Key information about MiniMax M2 and its available data.
MiniMax M2 is an open-source large language model by MiniMax, built for agents and coding tasks. 5 Sonnet’s cost and running at nearly double its inference speed (≈100 TPS). Designed for end-to-end agentic workflows, it excels at long-chain tool calling across Shell, Browser, Python, and other MCP tools.
While slightly behind top overseas models in programming, it ranks among the best domestic models and top five globally on the Artificial Analysis benchmark.
M2 powers the MiniMax Agent platform, available in Lightning Mode for fast tasks and Pro Mode for complex multi-step reasoning, and its weights, API, and deployment guides are freely available on Hugging Face, vLLM, and SGLang.
Data as of 2026-08-11.
MiniMax M2 vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to MiniMax M2
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
MI
MiniMax M1 80K
MiniMax
MI
MiniMax M1 40K
MiniMax
MI
MiniMax M2.1
MiniMax
MI
MiniMax M2.5
MiniMax
MI
MiniMax M2.7
MiniMax
MI
MiniMax M3
MiniMax
FAQ
Common questions about MiniMax M2.
When was MiniMax M2 released?
MiniMax M2's default version was released on Oct 27, 2025.
How much does MiniMax M2 cost?
MiniMax M2's official API price is $0.30 per million input tokens and $1.2 per million output tokens via MiniMax (minimax.io). The lowest tracked third-party offer starts at $0.17 input and $1.53 output via NanoGPT.
Who created MiniMax M2?
MiniMax M2 was created by MiniMax.
What is the context window for MiniMax M2?
The default version has a 196.6K token context window.
Is MiniMax M2 open weight?
Yes. The default version is marked as open weight under MIT.
How many API providers offer MiniMax M2?
17 provider offerings are linked to the default version.
What models should I compare MiniMax M2 with?
Nearby ranked alternatives include Sarvam 105B, DeepSeek-R1, GPT-OSS-120B.