MiniMaxMiniMax-M2
Best configuration ranks #182 of 610 on the BenchLeader Index at 55.1 ±3.1. Last measured 8 Sept 2026. Released 27 Oct 2025.
- Blended price
- $0.525/M
- $0.300 in · $1.20 out
- Output speed
- 88 tok/s
- First answer
- 25 s
- first token 0.93 s
- Context
- 205k
- Overall index55
- Reasoning55
- Coding51
- Agents & tools50
- Maths52
- Knowledge52
- Instruction following70
- Human preference56
- Long context60
- Composite55
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source | Trend |
|---|---|---|---|
| LMArena Hard Prompts | 1407#146 | LMArena | |
| GPQA Diamond (AA)not in index | 83.0%#159 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 23.2%#143 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | 78.5%#70 | Vals AI | |
| Kagi LLM Benchmark | 57.8%#57 | Kagi LLM Benchmark |
Coding
| Benchmark | default | Source | Trend |
|---|---|---|---|
| LMArena Coding | 1438#140 | LMArena | |
| LMArena WebDev | 1388#76 | LMArena | |
| LiveCodeBench | 81.8%#58 | Vals AI | |
| IOI | 2.3%#55 | Vals AI | |
| SWE-bench (Vals)not in index | 74.8%#45 | Vals AI | |
| SWE-bench Verified (bash only) | 61.0%#22 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in index | 61.0%#28 | SWE-bench |
Agents & tools
| Benchmark | default | Source | Trend |
|---|---|---|---|
| Terminal-Bench | 36.6%#39 | Terminal-Bench | |
| Terminal-Bench Hard | 28.8%#115 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 86.8%#82 | Artificial Analysis |
Maths
| Benchmark | default | Source | Trend |
|---|---|---|---|
| AIME (Vals) | 77.9%#49 | Vals AI | |
| MGSM | 87.8%#54 | Vals AI |
Knowledge
| Benchmark | default | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -32.9#224 | Artificial Analysis | |
| MMLU-Pro | 87.0%#35 | Vals AI | |
| LegalBench | 75.5%#105 | Vals AI | |
| CorpFin | 58.9%#79 | Vals AI | |
| TaxEval | 66.3%#109 | Vals AI | |
| MedQA | 91.2%#41 | Vals AI |
Instruction following
| Benchmark | default | Source | Trend |
|---|---|---|---|
| IFBench | 72.3%#49 | Artificial Analysis |
Human preference
| Benchmark | default | Source | Trend |
|---|---|---|---|
| LMArena Text | 1384#148 | LMArena |
Long context
| Benchmark | default | Source | Trend |
|---|---|---|---|
| AA-LCR | 67.7%#181 | Artificial Analysis |
Composite
| Benchmark | default | Source | Trend |
|---|---|---|---|
| AA Intelligence Index | 20.9#164 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Google Vertex | 122 tok/s | 0.45 s | $0.300 | $1.20 | 197k | – |
| NovitaAI | 57 tok/s | 1.28 s | $0.300 | $1.20 | 205k | fp8 |
| MiniMax | 56 tok/s | 0.93 s | $0.255 | $1.02 | 205k | fp8 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.030 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0005 | $0.0004 | 28.2 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0043 | $0.0019 | 31.6 s |
| Code edit | 6,000 / 1,500 | $0.0036 | $0.0024 | 41.8 s |
| Agentic coding session | 60,000 / 4,000 | $0.023 | $0.011 | 1.2 min |
| Structured extraction | 2,000 / 200 | $0.0008 | $0.0004 | 27.1 s |
See also
Data as of 9 Sept 2026. Compare with another model.