Mistral Large 3 2512
Neural Network
Mistral Large 3 2512 is a Mistral language model with sparse MoE architecture and Apache 2.0 license, available in BotHub via API.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 262 144 tokens
- Max answer length
- 209 715 tokens
- Capabilities
- function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Mistral Large 3 2512?
Mistral Large 3 2512 is a large language model by Mistral based on a sparse mixture-of-experts architecture: out of 675B parameters, 41B active ones are used per request, requiring fewer computations than a dense model of comparable size. The model is released under the Apache 2.0 license. In BotHub, the model connects without a VPN or foreign card, and you pay in rubles as you go — for consumed tokens that do not expire. With over 250 neural networks in one window, it's easy to compare responses to the same prompt, and a unified OpenAI-compatible API allows you to switch models by changing one parameter without rewriting the integration; chats are encrypted via AES-GCM, and companies have access to contracts, invoices, electronic document management, and an admin panel with limits. Mistral Large 3 2512 is typically used for code generation and refactoring, debugging third-party projects, drafting documentation and analytical notes, as the core of an internal assistant or chatbot, and for step-by-step analysis of math and science problems.Frequently asked questions about Mistral Large 3 2512
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 a Mistral text model: it accepts text, images, and documents as input and outputs text. It is suitable for working with code, analyzing files and long materials, analytics, as well as scenarios involving external tool calls. Context is 262,144 tokens.
Up to 209,715 tokens in a single response — enough for a large document, detailed instruction, or extensive refactoring without splitting the task. If you need a complete result, explicitly ask the model not to shorten it; in BotHub, you pay as you go for consumed tokens.
A separate reasoning mode for this model is not noted in our data, but that doesn't mean it doesn't exist. In practice, it helps to ask the model to break down the task step-by-step in the prompt itself: with a 262,144 token context, there is enough room for detailed thought processes.
Yes, both function calling and structured JSON output are confirmed. The model can call your tools and return responses in a specified schema. In BotHub, this works via a unified OpenAI-compatible API, so you can switch models without rewriting the integration.
Images and documents — yes: send a screenshot, diagram, PDF, or other file and get a text analysis. Audio and video input is not noted in our data, nor is the generation of anything other than text. For such tasks, BotHub has specialized models.
Language-specific characteristics are not recorded in our data, so we won't make specific promises — it is more reliable to test the model on your real task directly in BotHub, without a VPN or foreign card. If you are not satisfied with the result, there are over 250 other models in the same window.