Mistral Small 4
Neural Network
Mistral Small 4 is a Mistral AI model for reasoning, coding, and integration via the Mistral Small 4 API.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 262 144 tokens
- Max answer length
- 209 715 tokens
- Reasoning effort
- none, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Mistral Small 4?
Mistral Small 4 is another major release in the Small line from Mistral AI. Its main idea is unification: the capabilities of several flagship Mistral models are combined into one system. Its strength lies in reasoning—tasks where you need not just to continue text, but to build a logical chain and reach a conclusion. Compared to previous Small versions, this is not a targeted improvement of a single skill, but a consolidation of different competencies into a single framework. In BotHub, the model is accessible without a VPN or foreign card: pay with a Russian card in rubles only for tokens actually used; no subscription is needed, and unused balance does not expire. Over 250 models are available in the same window, allowing you to compare answers and switch to another neural network without rewriting your integration—the unified OpenAI-compatible API makes this possible. For teams, we offer contracts, invoices, electronic document management, and an admin panel with limits. Mistral Small 4 is useful for analyzing and refactoring code, debugging errors, preparing analytical reviews and document drafts, solving STEM-logic tasks, and acting as an assistant engine with function calling.Frequently asked questions about Mistral Small 4
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.
mistral-small-2603 is a Mistral text model with a 262,144 token context. It analyzes documents and images, reasons before answering, and calls functions. Suitable for coding: writing, refactoring, debugging, as well as analytics, long materials, and STEM tasks.
The single response limit is 209,715 tokens, which is very high for text models: enough for a voluminous document, a large code module, or a detailed analysis in full, without splitting into parts. In practice, request the desired volume directly in the prompt.
Yes, reasoning before answering is confirmed. The model breaks the task into steps and only then provides the result—this is noticeable in math, logic, and code analysis. In BotHub, you simply formulate the task, and the reasoning mode works on the model side.
Yes, the model supports function calling and structured JSON output. You describe the tools, it selects the necessary one and returns arguments in the specified schema—convenient for agents and integrations. Through BotHub's unified OpenAI-compatible API, switching to another model does not require rewriting code.
You can input images and documents in addition to text: the model will read a screenshot, diagram, table, or PDF and respond with text. There are no notes in our data regarding audio and video, so plan your work around text, images, and files.
The quality of Russian in our data is not noted, so it is better to check with your own material: give the model a typical request and evaluate the result. In BotHub, this is convenient—other models are available nearby, so you can compare answers in one window and choose the best one.