MiMo-V2.5 is the versatile, high-efficiency backbone of Xiaomi's 2.5-series AI ecosystem. While inheriting the "Harness-Awareness" architecture from the Pro version, MiMo-V2.5 is optimized for lower latency and extreme token efficiency in repetitive, high-frequency autonomous tasks. It excels at managing medium-to-long execution trajectories (up to 100+ tool calls) and maintaining precise context alignment. In comparative testing, it achieved a 50% reduction in token consumption compared to previous generation models like Muse Spark while maintaining superior task success rates. It is the ideal engine for 2026's decentralized AI micro-services and real-time coding assistants.
All providers = every upstream serving this model supports it. Some providers = depends on which upstream handles the request. Default = the value sent when you leave the parameter unset.
| Parameter | Providers | Default |
|---|---|---|
| frequency_penalty | Some providers | Not sent by default |
| include_reasoning | All providers | - |
| logit_bias | Some providers | - |
| logprobs | Some providers | - |
| max_tokens | All providers | - |
| min_p | Some providers | - |
| presence_penalty | Some providers | Not sent by default |
| reasoning | All providers | - |
| repetition_penalty | Some providers | Not sent by default |
| response_format | Some providers | - |
| seed | Some providers | - |
| stop | Some providers | - |
| structured_outputs | Some providers | - |
| temperature | All providers | 1 |
| tool_choice | All providers | - |
| tools | All providers | - |
| top_k | Some providers | Not sent by default |
| top_logprobs | Some providers | - |
| top_p | All providers | 0.95 |