Xiaomi MiMo v2.5 and v2.5 Pro in the Yenisei catalog

We have added the Xiaomi MiMo lineup: the base MiMo v2.5 (mimo-v2.5) and the senior MiMo v2.5 Pro (mimo-v2.5-pro). Xiaomi rarely comes up in conversations about language models, but the lineup turned out serious, and in places very interesting in terms of price to quality.

What is inside

• MiMo v2.5 is a solid workhorse for general tasks: dialogue, texts, simple code.
• MiMo v2.5 Pro is the version with reasoning, tool calling (tools) and a context of up to 1M tokens.

How to call

curl https://api.yenisei.ru/v1/chat/completions \
  -H "Authorization: Bearer $YENISEI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mimo-v2.5-pro",
    "messages": [
      {"role": "user", "content": "Analyze this stack trace and suggest where to look for the cause."}
    ]
  }'

Who should be interested

MiMo v2.5 Pro is worth trying if you need reasoning and a large context, but not at the price of the most expensive flagships. The combination "a million tokens plus reasoning plus tools" is rare in this price segment. Take the base v2.5 for the general stream where neither deep deliberation nor a very long context is needed.

One caveat: on the hardest reasoning tasks the best closed models are still ahead. But in many practical cases MiMo v2.5 Pro gets the job done and costs noticeably less; the savings are worth checking on your own tasks.

Both models handle Russian fine. Prices are in rubles, 1 credit = 1 ₽. Both are already in the catalog.