Pixtral Large 2502
Neural Network
Pixtral Large 2502 is a Mistral AI multimodal model that analyzes documents, charts, images, and text in a single request
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 128 000 tokens
- Max answer length
- 102 400 tokens
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
import OpenAI from 'openai';
const openai = new OpenAI({
apiKey: '<your bothub access token>',
baseURL: 'https://openai.bothub.chat/v1'
});
// Sync - Text generation
async function main() {
const chatCompletion = await openai.chat.completions.create({
messages: [{ role: 'user', content: 'Say this is a test' }],
model: 'pixtral-large-2502',
});
}
// Async - Text generation
async function main() {
const stream = await openai.chat.completions.create({
messages: [{ role: 'user', content: 'Say this is a test' }],
model: 'pixtral-large-2502',
stream: true
});
for await (const chunk of stream) {
const part: string | null = chunk.choices[0].delta?.content ?? null;
}
}
main();
How it works Pixtral Large 2502?
Pixtral Large 2502 is the February version of Mistral AI's large multimodal model, continuing the Pixtral Large 2411 line. It accepts not only text but also images: documents, diagrams, charts, and regular photos, allowing a single request to combine a verbal task and an image. In BotHub, the model is accessible without a VPN or foreign card; you pay with Russian cards only for actual tokens used, and the balance does not expire. You can work via the web interface or Telegram bot, switching between Pixtral Large 2502 and other models in the catalog within one window. For your services, you can connect a single OpenAI-compatible API without rewriting integrations when switching models. Typical tasks include analyzing photos of documents and diagrams, reading charts in reports, describing images for product catalogs, checking interface screenshots, and verifying data from images against text. Teams have access to contracts, invoices, electronic document management (EDO), and limits in the admin panel, with data transmission protected by AES-GCM encryption.Frequently asked questions about Pixtral Large 2502
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.
Pixtral-large-2502 is a Mistral multimodal model: it accepts text, images, and documents, and responds with text. It is useful when you need to analyze scans, tables, charts, or screenshots and get a coherent description, summary, or answer to a question. The context is 128,000 tokens, so a large document can fit into a single request along with your comments.
In a single response, the model can output up to 102,400 tokens — this is the upper limit based on our data, not a guarantee for every request. This volume is sufficient for long analyses, detailed image descriptions, and entire documents. The actual response length depends on your prompt and how much space is left in the context.
A separate reasoning mode for pixtral-large-2502 is not noted in our data, so we cannot confirm it: the model answers immediately, without a separate thought block. For multi-step tasks, it helps to break the question into parts and ask the model to show the solution process in the response itself — this makes the result easier to verify.
Function calling and structured JSON output are not noted in our data, so we will not promise them — please check the Mistral documentation for the current set of capabilities. If you need a stable JSON response, request the desired format directly in the prompt and verify the result on your end.
Yes, in addition to text, the model accepts images and documents — this is confirmed in our data. You can upload a scan, photo, table, or PDF and ask a question about the content. Audio and video are not noted in the data, so it is better not to count on uploading such files: another model in BotHub will be suitable for them.
There is no separate quality assessment for the Russian language in our data, so we will not promise a specific level. The easiest way to check is with your own tasks: upload a typical document or scan and see how accurately the model parses the wording and terms. Access to pixtral-large-2502 is provided via BotHub — without a VPN or foreign card.