Codestral 2508
Neural Network
Codestral 2508 is a Mistral model for code: fast auto-completion (FIM), bug fixing, and test generation.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Codestral 2508?
Codestral 2508 is a Mistral language model designed specifically for coding, released in late July 2025. It specializes in low-latency, high-frequency tasks: fill-in-the-middle code completion, bug fixing in existing snippets, and test generation, making it ideal for editors and build pipelines where sub-second responses are required. In BotHub, Codestral is accessible from Russia without a VPN or foreign card: you pay in rubles for tokens used, and unused tokens do not expire. Over 250 other models are available in the same window, allowing you to compare responses or switch with a few clicks. A unified OpenAI-compatible API lets you swap models in your multi-model service without rewriting integrations, and data is protected by AES-GCM encryption. For teams, we offer contracts, invoicing, EDI, and an admin panel with limits. Typical scenarios include auto-completion within existing files, analyzing and fixing failing code, wrapping functions with unit tests, accelerating reviews, and refactoring legacy modules.Frequently asked questions about Codestral 2508
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.
Codestral-2508 by Mistral is a model for coding: writing functions, refactoring, explaining unfamiliar snippets, debugging, and troubleshooting. The 256,000-token context allows you to load a large module or documentation in its entirety and discuss it without losing details.
Up to 204,800 tokens in a single response — enough for an entire module with tests or a detailed architectural analysis. In practice, it is more convenient to request in parts: this makes it easier to verify results and make edits without regenerating the entire volume.
A separate reasoning mode is not noted in our data for codestral-2508, so do not expect a visible chain of thought. However, you can ask the model to break down the task step-by-step directly in the prompt — this usually works for logic analysis and debugging.
Yes, function calling and structured JSON output are supported. The model can be connected to external tools, databases, and custom services, and you can receive responses in a specified schema — convenient for agents and automation. In BotHub, this is available via a unified OpenAI-compatible API.
Codestral-2508 accepts text and documents as input — you can attach a code file, specification, or technical documentation. Images, audio, and video are not supported, but in BotHub, you can switch to a multimodal model in the same window and via the same API.
The quality of Russian in the model itself is not described in our data, so we will not promise precise evaluations — please test it with your own tasks. The BotHub interface is in Russian, payment is in rubles, access is without VPN or foreign cards, and you can compare the response with another model on the same prompt in one window.