How to Use Mistral Nemo on Composite
At a glance
| Model ID | mistralai/mistral-nemo |
|---|---|
| Free ID | mistralai/mistral-nemo:free |
| Made by | Mistral AI with NVIDIA |
| Context window | 131,072 tokens |
| Longest reply | 16,384 tokens |
| Input | Text |
| Reasoning | No |
| Tool calling | Yes |
Mistral Nemo is a 12B-parameter model that Mistral AI built with NVIDIA, released in July 2024. It has a 128k-token window and is multilingual, with support for English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese and other languages. It was released with open weights, and many community roleplay fine-tunes start from it.
Use a low temperature
Mistral's own model card says Nemo needs lower temperatures than its earlier models and recommends 0.3. Roleplay presets often default to 0.8 or higher, which tends to make Nemo ramble or lose track of the scene. Start at 0.3 to 0.5 and raise it only if replies feel too predictable.
Settings it accepts
Nemo takes the full chat sampler set: temperature, top P, top K, min P, repetition penalty, and frequency and presence penalties, plus stop sequences, logit bias and a JSON response format. It also supports tool calling. It has no reasoning mode, so do not add an effort suffix such as [high] to the model ID.
Writing in other languages
Because the model was trained for several languages, it is a sensible free choice for chats that are not in English. Write the character card and the first message in the language you want; the model tends to follow the language of the prompt. If it drifts back to English, add one line to the system prompt naming the language of every reply.
Context and reply length
The 131,072-token window fits a long chat with a full preset and lorebook. Replies can be up to 16,384 tokens, so set your app's response length well below that for chat. When a chat outgrows the window, Composite trims the oldest messages to fit.
Connect it to your chat app
- Create an API key on the API Keys page.
- In SillyTavern, ChubAI, Agnai or RisuAI, choose an OpenAI-compatible provider and set the base URL to
https://composite.lucidity.sh/v1. JanitorAI wants the full URL:https://composite.lucidity.sh/v1/chat/completions. - Paste the key, then type
mistralai/mistral-nemo:freeas the model name. - Send a short message to check the connection before you load a long chat.
Step-by-step setup for each app is in the JanitorAI and SillyTavern guides, and the docs hub covers the others.
From code, it is a normal OpenAI-compatible request:
curl https://composite.lucidity.sh/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "mistralai/mistral-nemo:free", "messages": [{"role": "user", "content": "Hello"}]}'Free access (as of October 21, 2026)
On October 21, 2026, this model is on Composite's free list as mistralai/mistral-nemo:free. The plain mistralai/mistral-nemo stays paid. Every account shares 80 requests a day across all of the free models, separate from Synth's count, and the count resets at midnight UTC. Credits and plans do not raise it.
The free list changes. Check the free models on the Models page for the current set, and the features guide for the rules. A :free id that is no longer on the list returns a model-not-found error and charges nothing. When the 80 requests are used up, requests stop with an error until the reset; Composite never charges your credits to continue. You can switch to mistralai/mistral-nemo on credits or a plan, or to Synth.