Free AI playground
Chat with Mistral Large 4 online
Review code, draft text, or ask about an image right in your browser. Sign in for 30 free chat requests every day. No API key needed.
- Model context window
- 1M
- Playground
- Free
- Input types: text and images
- 2
Mistral Large 4 playground
Test your next prompt
Chat with Mistral Large 4. Choose a reasoning mode, request JSON, or attach a public image URL.
Use Playground for free
Mistral Large 4
Model API · text and image input
What would you like to explore?
Ask a question, review code, work through a document, or attach a public image URL.
Try a field test
What is Mistral Large 4?
Mistral Large 4 is a general-purpose multimodal AI model. It works with text and images for tasks such as coding, document questions and tool-based workflows.
Information checked October 7, 2026
How can I try it here?
Open the playground and sign in to use it for free. Ask a question, review code or add an image URL. When you are ready to integrate, create an account API key and follow the step-by-step documentation.
Try playgroundIs it right for your work?
Start with a task you can verify: explain a function, summarize a passage with evidence, or inspect a screenshot. Compare the answer, response time and cost against your current tool.
Compare models by taskHow do I use Mistral Large 4 online?
No coding is needed for the playground. Bring one task, try an answer, and refine it with a follow-up.
Sign in and open the playground
Sign in to use the playground for free. Enter a prompt or choose an example to start your first conversation.
Open the playgroundAdd the context that matters
Paste text or code, or attach a public image URL. State the task and desired output. For a PDF, extract the relevant text first.
See prompting tipsCheck the answer and follow up
Ask for evidence, run suggested code, and correct missing context. Review credits and plans when you need more usage.
Understand credits and plans
Capabilities in practice
What can you do with Mistral Large 4?
Use Mistral Large 4 to review code, reason over supplied text and interpret images. The task examples below describe what to try and what to check.
1M
Model context window
Available input size depends on the service you use.
Ask questions about long documents
Mistral Large 4 supports a 1M-token context window. This playground accepts shorter excerpts and public image URLs, not PDF uploads. Ask for evidence from the text you provide.
Review and explain code
Share the relevant functions, error messages and expected behavior. Ask for a diagnosis and a small patch, then run your own checks before adopting it.
- text
- image
- json
- tools
Understand screenshots and diagrams
Attach a public image URL with a specific question. Get a text explanation, and check small labels, numbers and visual details against the original.
tools = [get_file, search_records] response = chat({ model, messages, tools }) await execute_approved(response.tool_calls)
Build an assistant that uses tools
Through the API, define functions for search or data lookup. The model proposes tool calls; your application validates and executes them.
$ curl -X POST https://mistrallarge4.com/api/v1/chat/completions -H "Authorization: Bearer sk_…" 200 OK
Turn answers into structured data
Request JSON for classification, extraction or a next-step workflow. Validate the result against your schema before passing it to other systems.
Can it help with writing and research?
Draft, rewrite or summarize content for your audience. Add your source material, preferred language and tone, then ask for supporting passages when accuracy matters.
Is Mistral Large 4 free? What does it cost?
Use the playground for free. When you need more usage, choose a credit pack for occasional use or a monthly subscription for ongoing work.
Use Playground for free
Sign in to use the playground for free. Additional usage beyond the free allowance consumes credits: actual model cost in USD × 12,000, rounded up. Requests admitted for generation count toward the free allowance even if later cancelled or failed.
Credit example: a completed answer with $0.01 of model usage consumes 120 credits. The dollar price of those credits depends on the pack or subscription you buy.
View credit plansChoose a plan for your usage
One-time credit packs do not auto-renew and their credits never expire. Monthly subscriptions grant credits each billing period, with each grant valid for 90 days. Compare the available plans on the pricing page.
For occasional projects, start with a credit pack. For regular use, compare monthly plans against your actual credit consumption. Your account keeps the balance and billing together.
Compare monthly plansFree access is available in the playground. Review usage details in the documentation and manage plans in your account.
Put it to work
A long-context model for more than chat
Mistral Large 4 combines coding, multimodal input, tool calling and adjustable reasoning in one model endpoint.
response = client.chat.completions.create(
model="mistral-large-4-0",
messages=[{"role": "user", "content": repo_context}],
extra_body={"reasoning": {"effort": "high"}},
)
review = response.choices[0].message.content// trace
- context_window524,288
- reasoning_efforthigh
Designed for developers
How do I connect Mistral Large 4 to my app?
Create an API key in your account, then use the base URL and model ID together. Follow the Python example below to send your first request. Start with text, then add images, JSON or tools to suit your workflow.
- OpenAI-compatible REST
- Streaming, JSON and tools
- Server-side API keys
import os
from openai import OpenAI
client = OpenAI(
base_url="https://mistrallarge4.com/api/v1",
api_key=os.environ["MODEL_API_KEY"],
)
response = client.chat.completions.create(
model="mistral-large-4-0",
messages=[{"role": "user", "content": "Review this design"}],
max_tokens=2048,
extra_body={"reasoning": {"effort": "none"}},
)
print(response.choices[0].message.content)A minimal request: Account API key · Chat endpoint
# Configure an OpenAI-compatible client
# 1. endpoint
export OPENAI_BASE_URL="https://mistrallarge4.com/api/v1"
# 2. configure
export OPENAI_API_KEY="sk_your_key_here"
export MODEL_ID="mistral-large-4-0"Your application keeps control
The model can request a tool call, but your code still validates permissions, arguments, side effects and human approval boundaries.
Agent integration
Give your agent more room to think
Connect Mistral Large 4 to your agent stack and choose the reasoning effort for each task.
- 1
Connect to the API endpoint
Point your client at the API base URL and select mistral-large-4-0. Create your key in account settings.
- 2
Keep the key server-side
Store your account API key in the server environment. Never expose it in browser code or public prompts.
- 3
Offer only approved tools
Define the functions your agent may request, then validate arguments and permissions before execution.
Practical guides
Before you build with Mistral Large 4
Follow the question you are trying to answer: how to connect, what it costs, which model to choose, or when self-hosting becomes possible.
How do I call the API?
Create an API key and follow Python examples for streaming, image input and tool calling.
What will I pay?
Learn about free playground access, usage-based credits, one-time packs and monthly subscriptions.
Which model should I choose?
Compare Large 4 with Large 3 and evaluate alternatives for coding, document analysis and agents.
Can I download and deploy it?
Check the weight-release plan, licensing questions and what to verify before sizing hardware.
Recent blog posts
Tutorials and model updates
Prompt tutorials, API integration examples and setup guides for Mistral Large 4.

Uncategorized
Mistral Large 4: What “Le Chonk” Means for Developers and AI Teams
A practical, source-aware guide to Mistral Large 4, also known as Le Chonk: its 1.05T-parameter MoE design, 1M context window, multimodal and agentic capabilities, preview API, open-weight roadmap, and production trade-offs.
Oct 30, 2026 · 1 min read

Uncategorized
What Is Mistral Large 4? A Practical Guide to the Multimodal Model
What is Mistral Large 4? This practical guide explains its MoE architecture, multimodal vision, long context, tool calling, structured output, API model ID, pricing notes, and limitations.
Oct 7, 2026 · 1 min read
Plans and usage
Mistral Large 4 Pricing
Use the playground for free after signing in. Choose a one-time credit pack or a monthly subscription when you need more, and manage usage and billing in your account.
One-time credit packs never expire and do not auto-renew. Monthly subscriptions grant credits each billing period, valid for 3 months (90 days) from each grant.
Your questions about Mistral Large 4
Straight answers about using Mistral Large 4, choosing a model and planning an integration. Model-release information checked October 7, 2026.
Mistral Large 4 is a multimodal model for text and image tasks. Use the playground to try prompts, create account API keys for integrations, and choose credit plans for continued use.
Read moreYes. Sign in and use the playground for free — no download, server or GPU needed. When you need more than the free allowance, credits are charged by actual model usage: model cost in USD × 12,000, rounded up. Pick a credit pack or a monthly plan on the pricing page.
Read moreThe playground is free after sign-in. Beyond the free allowance you pay with credits: buy a one-time credit pack that never expires, or a monthly plan for regular use. The same balance covers playground chats and API requests made with your account key.
Read moreOpen the playground and sign in to use it for free. Paste a prompt or select an example to test text, code and public image URLs without downloading weights or setting up a server.
Read moreThere is no single winner for every task. Compare a specific model version, the same source material, equivalent tools and a fixed budget. Check correctness, evidence, latency and total task cost. A chat product’s features and subscriptions are separate from the underlying model.
Read moreIt handles code review, debugging, refactoring and explaining unfamiliar code, and supports tool calls for agent workflows. The best test is your own code: paste a real bug into the playground and check whether the suggested patch is correct and minimal.
Read moreLarge 3 handles 256k tokens of context; Large 4 supports up to 1M. Large 4 is the newer generation and still evolving, so re-run your real tasks and compare reliability, speed and credit usage before migrating. A larger context alone does not guarantee better results.
Read moreMistral Large 4 supports a 1M-token context window. Our API integration is configured around 524,288 tokens, and the playground has a smaller 96 KiB request limit, 24 messages and up to 16,384 output tokens. Model capacity is not a promise that every interface accepts that much input.
Read moreCreate an account API key and follow the integration guide. Use the base URL with model ID mistral-large-4-0. Start with a short text request, then add streaming, images, structured JSON or tool calls as your application needs them.
Read moreYes. Point the OpenAI SDK at our base URL with your account API key and model ID mistral-large-4-0. Chat completions support streaming, image URLs, tool (function) calling and JSON schema structured output.
Read moreNot on a laptop or a single consumer GPU. At roughly trillion-parameter scale, the weights alone take about 0.5 TB even at 4-bit precision, which means a multi-GPU server. Use the hosted playground and API here to get the model without buying hardware.
Read moreAs of October 7, 2026, Mistral Large 4 is available as a hosted model and the weights are expected later. Running it yourself will need multi-GPU server hardware. Until then, the playground and API here give you the same model with nothing to install.
Read moreThis playground exposes none for a direct answer and high for deeper reasoning. Reasoning uses part of the output budget. If a response is cut short, increase the output limit within the allowed range or narrow the task.
Read moreThis playground accepts pasted text and public image URLs, then returns text. Direct PDF uploads, audio and video inputs are not supported here. Extract relevant PDF text first; an image URL can be used for a page or chart that needs visual interpretation.
Read moreYes. Paste your source material and specify the audience, language and output format. Ask for a draft, a rewrite or a summary with supporting passages. Check quotations and numbers against the original before publishing.
Read moreRequests are processed through a hosted model provider. Before using confidential material, review the privacy policy and confirm retention, training use, processing region and contractual requirements with support. The playground is not a private deployment.
Read moreBring a task you can check.
Try Mistral Large 4 with code, a document excerpt or a screenshot. Use the playground for free, then bring the same workflow into your app with the API.
Open the playground