MiMo v2.5 vs MiMo v2.6 Pro
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
MiMo v2.527.2
MiMo v2.6 Pro49.4
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
MiMo v2.50
MiMo v2.6 Pro34.8
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
MiMo v2.518.4
MiMo v2.6 Pro58.6
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
MiMo v2.51,395
MiMo v2.6 Pro1,445
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
MiMo v2.575.4
MiMo v2.6 Pro—
| Benchmark | MiMo v2.5 | MiMo v2.6 Pro |
|---|---|---|
| Humanity's Last Exam | 27.2 | 49.4 |
| Terminal-Bench 4.0 | 0 | 34.8 |
| AutomationBench-AA | 18.4 | 58.6 |
| Arena Text · creative writing | 1,395 | 1,445 |
| MMMU-Pro (vision) | 75.4 | — |
Specs
| Spec | MiMo v2.5 | MiMo v2.6 Pro |
|---|---|---|
| Input, per 1M | $0.14 | $0.43 |
| Output, per 1M | $0.28 | $0.87 |
| Cost per task? | $0.019 | $0.13 |
| Speed, tok/s | 38 | 54 |
| Context | 1M | 1M |
| Weights | Open | Open |
| Released | Apr 2026 | Sep 2026 |
| Memory (4-bit) | 186 GB | 615 GB |
| Licence | MIT | MIT |
| Commercial use | Allowed | Allowed |
| See images | ✓ | ✓ |
| Read PDFs | — | — |
| Hear audio | ✓ | ✓ |
| Watch video | ✓ | ✓ |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ? |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
MiMo v2.5 → MiMo v2.6 Pro
Memory (4-bit): 186 GB → 615 GB
Pass back reasoning_content›
In deep-thinking multi-turn tool conversations, return the full reasoning_content in later rounds with tool calls or receive HTTP 400.
Forced recommended defaults›
Deep thinking overrides supplied temperature and top_p with 1.0 and 0.95.
Avoid answer truncation›
max_completion_tokens includes thinking and the final answer; reserve enough tokens to avoid truncating the answer.
MOPD checkpoints address tool-call repetition›
Use the MOPD checkpoints for the tool-call repetition fix; the API has served the updated models since September 25 at 06:00 UTC+8 under unchanged names.
MiMo v2.6 Pro → MiMo v2.5
Nothing known.