X
MiMo-V2.5OPEN
MiMo-V2.5 is a native omnimodal model by Xiaomi.
Prices update automatically — checked daily against provider list prices.
- $0.18 per 1M tokens, assuming 3 input for 1 output the convention used by Artificial Analysis
- cheaper than 64% of the 191 open-weight models tracked here
Benchmarks & performance
Independent benchmark scores for MiMo-V2.5, measured by Artificial Analysis. Higher is better.
For the same budget, MiMo-V2.5 delivers 34 times more capability. Open weights also mean you can self-host it and pay nothing per token.
Models at this level cost $1.56 per 1M tokens (median of 24) — this one costs $0.18.
Cheaper and better on this index: GLM 5.3 Flash · DeepSeek V4 Flash 0731 · DeepSeek V4 Flash 0423 and 2 more
About this model
MiMo-V2.5 is an open-weight AI model by Xiaomi. You can download and self-host it for free; the prices below are hosted-API list prices, tracked daily, for when you prefer convenience over self-hosting.
Frequently asked questions
What is MiMo-V2.5?
MiMo-V2.5 is an AI language model from Xiaomi. It is open-weight: you can download it and run it on your own hardware, for free. It scores 38 on the Artificial Analysis intelligence index.
Is MiMo-V2.5 free?
The weights are free and open — you can self-host MiMo-V2.5 and pay nothing per token. If you prefer a hosted API, list prices are $0.14 per million input tokens and $0.28 per million output tokens.
What is MiMo-V2.5 good at?
Independent benchmarks from Artificial Analysis give it GPQA 84.9%, Humanity's Last Exam 27.2%, Long Context Reasoning 68.3%, SciCode 43.1%, IFBench 67.1%, τ²-Bench 90.6%, τ-Bench Banking 8.7%, Terminal-Bench 63.7%, Terminal-Bench Hard 41.7%. It is particularly used for code generation.
How fast is MiMo-V2.5?
It generates about 67.2 tokens per second, with a median 2.01s delay before the first token. Measured independently by Artificial Analysis.
Can I self-host MiMo-V2.5?
Yes. MiMo-V2.5 has open weights, so you can download it and run it on your own GPU or server with tools like Ollama, vLLM or llama.cpp — with no per-token cost.