Pyyan / Compare / gpt-oss-120b vs MiMo-V2.6-Pro
gpt-oss-120b vs MiMo-V2.6-Pro
2 of 5Open-Weight Models · verified 29 Sept 2026
gpt-oss-120bOpenAIcurrentXMiMo-V2.6-ProXiaomicurrent
| Specification | gpt-oss-120b | MiMo-V2.6-Pro |
|---|---|---|
| Summary | OpenAI's open-weight release. | A trillion parameters under the MIT licence. |
| Context | — | 1M |
| Input ($/Mtok) | — | Weights only |
| Output ($/Mtok) | — | Weights only |
| Licence | Open weights | MIT |
| Max output | — | Not published |
| Parameters | 120B | 1.02T total, 42B active |
| Released | Aug 2025 | 21 Sep 2026 |
| Model ID | — | XiaomiMiMo/MiMo-V2.6-Pro |
| Category | Open-Weight Models | Open-Weight Models |
| Official | OpenAI ↗ | Xiaomi ↗ |
Highlighted rows are where these differ.
MiMo-V2.6-Pro
- 1.02T total parameters, 42B active per token, a 4.1% activation ratio
- Text, image, video and audio in, with a 1M token context
- Shipped with 7,000+ reinforcement learning environments and the framework that trained it
Best for multimodal work on your own cluster.