Mistral Small 3.2 24B Instruct
Neural Network
Mistral Small 3.2 24B Instruct is an updated Mistral model for precise instruction following and API function calling.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Mistral Small 3.2 24B Instruct?
Mistral Small 3.2 24B Instruct is an updated 24-billion parameter model from the French lab Mistral, built around precise instruction following. Compared to version 3.1, it executes tasks more accurately, repeats itself less, and avoids looped responses. Function calling is more reliable, allowing the model to handle tool chains smoothly. Via BotHub, it is accessible without a VPN or foreign card: pay in rubles only for tokens used, not for a subscription, and unused tokens do not expire. With over 250 models in one window, you can compare responses to the same prompt. A unified OpenAI-compatible API lets you connect Mistral Small 3.2 to your service and swap it for another model later without rewriting the integration. For companies, we offer contracts, invoices, EDI, and an admin panel with limits. It is useful for building agents with external function calls, parsing documents into specific templates, returning answers in required formats, processing streams of similar requests or emails, drafting texts based on clear briefs, or embedding the model into products for high-volume requests.Frequently asked questions about Mistral Small 3.2 24B Instruct
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
This is an instruction-tuned Mistral model with 24 billion parameters. It parses documents and images, writes and edits code, answers questions, and prepares summaries, emails, and descriptions. The 32,768-token context allows you to work with an entire file or a long conversation at once.
The model generates up to 16,384 tokens per response — enough for a long article, detailed analysis, or a large code module. Need more? Ask it to continue: the 32,768-token context can fit both your prompt and the text already written.
A dedicated reasoning mode is not specified in our data, so do not count on it in advance. However, you can guide the thought process yourself: ask it to break down the task step-by-step, list conditions, and verify the result — this significantly helps with logic and code analysis.
Yes, function calling and structured JSON output are supported. The model is suitable for agents, parsers, and scenarios where the response goes directly to a database or third-party service. Connection is via the unified OpenAI-compatible BotHub API, so you won't have to rewrite the integration for another model.
You can upload images and documents: the model will read scans, tables, diagrams, or screenshots and respond with text. Audio and video are not among the supported input types — for such files, BotHub has separate models available in the same window and with the same payment method.
We do not provide specific language benchmarks in the model data, so it is most reliable to test it on your own task — usually, a couple of prompts are enough. If you are not satisfied with the result, there are over 250 other models in the same window: switch without a VPN and pay with a Russian card.