Models·Mistral

Mistral Nemo: What It Does and What It Costs

mistralai/mistral-nemo

Mistral’s open-weight text model supports tools and handles up to 131,072 tokens of input.

$0.019Input, per million tokens
$0.030Output, per million tokens
131KContext window
16KLongest answer

131K tokens of context — about 98,304 words, on a scale that ends at 2M.

What it can do

  • Tool use
  • Open weights
  • Takes text
  • Gives text

Released 19 Jul 2024 · checked 3 Oct 2026.

About it

Mistral Nemo is a 12-billion-parameter text model made by Mistral with NVIDIA. It accepts text, can use tools, and supports languages including English, French, German, Spanish, Italian, Portuguese, Chinese, and Japanese. Its input and output prices are low, and it has open weights. It can read up to 131,072 tokens at once and write up to 16,384 tokens in one answer.

Review

In a line

Mistral Nemo is a text model made by Mistral in collaboration with NVIDIA. It has 12 billion parameters, and its weights are open.

What it's good at

It can work with text in several languages, including English, French, German, Spanish, Italian, Portuguese, Chinese, and Japanese.

The model supports tools. It produces text, not images.

Price in plain terms

Input costs $0.019 per million tokens, and output costs $0.03 per million tokens. A page of about 500 words costs under 0.01 cents to read. One chat turn with about 1,000 tokens in and 500 out costs under 0.01 cents; 1,000 such turns cost about $0.03.

Those figures are token-use costs, not a flat subscription price.

Context and limits

It can take in up to 131,072 tokens at once. That is the amount of text it can consider together when responding.

A single answer can contain up to 16,384 tokens.

Ways to use it

Mistral Nemo is available as an open-weight model. People can use it through a service or app that offers the model; the facts do not name particular apps or integrations.

Its listed model ID is mistralai/mistral-nemo. See other Mistral models and models that support tool calling.

Verdict

Consider Mistral Nemo if you want an open-weight text model with multilingual support, tool use, and low per-token costs. Its large input limit may suit work that needs a lot of text considered together, while the output cap sets a clear ceiling on answer length.

Look elsewhere if you need a model that handles images, or if you need confirmed compatibility with a particular app: the facts list text only and do not name integrations.

Verdict

Mistral Nemo is worth considering for readers comparing low-cost text models, especially if open weights, multilingual text, or tool support matter. Its input and output rates are priced separately, and the stated cost for 1,000 typical chat turns is about $0.03. The large context window can accommodate substantial input, though answers have a defined maximum length. It is a poor fit for image work, and readers who depend on a specific app connection should confirm that the app offers it.

  • Open weights are available.
  • Supports English, French, German, Spanish, and other listed languages.
  • Can work with tools.
  • Input and output prices are low; 1,000 typical chat turns cost about $0.03.
  • Can take up to 131,072 tokens of input.
  • Accepts and produces text only.
  • Answers are limited to 16,384 tokens.
  • No particular apps or integrations are named.
  • Pick it ifPeople comparing low-cost text models; Tasks involving several supported languages; Use cases that need tool support or open weights
  • Skip it ifYou need image input or image generation.; You need a confirmed connection to a specific app.

Pricing

There is no flat plan listed. Input costs $0.019 per million tokens and output costs $0.03 per million tokens. The stated estimate for 1,000 chat turns is about $0.03.

Questions

How much does Mistral Nemo cost?

Input costs $0.019 per million tokens, and output costs $0.03 per million tokens. About 1,000 chat turns with 1,000 tokens in and 500 out each cost about $0.03.

What can Mistral Nemo read at once?

Its context window is 131,072 tokens, so it can consider that much text together.

How long can its answers be?

The maximum output is 16,384 tokens in one answer.

Can Mistral Nemo handle images?

No image support is listed. The model accepts text and produces text.

Does Mistral Nemo support tools?

Yes. It supports tools.

Which languages does it support?

Listed languages include English, French, German, Spanish, Italian, Portuguese, Chinese, and Japanese.

Is Mistral Nemo open source?

Its weights are open. The model is listed as mistralai/mistral-nemo.

On these boards

More from Mistral