M2.5
MiniMax · released Feb 12, 2026
MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digital working environments, M2.5 builds upon the coding expertise of M2.1...
Specification
- Context window
- 205K
- Max output
- 197K
- Knowledge cutoff
- Not stated
- Parameters
- 228.7B
- Licence
- other
- Serving providers
- 9
- Moderated
- No
- Uptime
- 100.0%
Intelligence
34.5
80th percentile
Coding
—
Coding Index
Agentic
—
Agentic Index
Output speed
91 t/s
Median across providers
Latency
549ms
Time to first token
Cost per task
$0.08
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- τ²-bench (Telecom)95.3%
- GPQA Diamond84.8%
- AA-LCR (long context)72.0%
- IFBench71.6%
- SciCode42.6%
- Terminal-Bench Hard34.8%
- Humanity's Last Exam20.5%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| τ²-bench (Telecom) | 95.3% |
| GPQA Diamond | 84.8% |
| AA-LCR (long context) | 72.0% |
| IFBench | 71.6% |
| SciCode | 42.6% |
| Terminal-Bench Hard | 34.8% |
| Humanity's Last Exam | 20.5% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- GLM 5V TurboZ.ai35.3
- Qwen3.5-27BQwen34.6
- M2.5MiniMax34.5
- GLM 4.7Z.ai34.5
- Hy3 previewTencent34.4
- 34.3
- LongCat 2.0Meituan33.9
- KAT-Coder-Pro V2KwaiPilot33.9
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| GLM 5V Turbo | 35.3 |
| Qwen3.5-27B | 34.6 |
| M2.5 | 34.5 |
| GLM 4.7 | 34.5 |
| Hy3 preview | 34.4 |
| Qwen3.5 397B A17B | 34.3 |
| LongCat 2.0 | 33.9 |
| KAT-Coder-Pro V2 | 33.9 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $0.22
- Output / 1M tokens
- $0.9
- Cached input / 1M
- $0.05
- Blended 3:1
- $0.39
One task, estimated
$0.08
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.