# Mistral’s February Le Chat refresh joined mobile access, code execution and enterprise preview

Mistral refreshed Le Chat on February 6, 2025, adding mobile access, a preview code interpreter, image generation, paid tiers and an Enterprise private preview.

Canonical: https://rangoon.ai/insights/archive-mistral-all-new-le-chat-february-2025/
Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/)
Published: 2026-10-24
Updated: 2026-10-24
Event date: 2025-02-06
Tags: [Mistral AI](https://rangoon.ai/insights/tags/mistral-ai/), [Le Chat](https://rangoon.ai/insights/tags/le-chat/), [consumer AI](https://rangoon.ai/insights/tags/consumer-ai/), [code interpreter](https://rangoon.ai/insights/tags/code-interpreter/)
Source note: Historical reporting: Mistral announced an all-new Le Chat on February 6, 2025. Its product post described iOS and Android access, Flash Answers preview, web and document features, a sandboxed code interpreter, Flux Ultra image generation, Pro and Team tiers, and Le Chat Enterprise in private preview. A November 2024 Mistral post gives prior beta context for search, Canvas, Pixtral Large and Flux Pro; it does not turn those earlier releases into February features or establish current availability.

Image: Multi-tool assistant evidence boundary. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied.

## Key takeaways
- The February release combined a consumer assistant refresh with paid plans and an Enterprise private-preview path.
- Code execution, image generation, web results and document handling are separate tool surfaces with separate evidence and data boundaries.
- Flash Answers, Memories and future connectors carried preview or rollout language that should not be reported as universal access.

## The February event was a product-surface expansion, not a model release [source 1](https://mistral.ai/news/all-new-le-chat/)

On February 6, 2025, Mistral presented an all-new Le Chat as an assistant for personal and work tasks. The company said it was available on iOS and Android, introduced Pro and Team service tiers, and put an Enterprise version in private preview. That combination made the event more than a new model announcement. It connected a consumer interface, mobile access, subscription choices and an enterprise evaluation path, each of which can have a different availability state and operational contract.
Mistral also described Flash Answers, claiming very high token throughput, as available in preview. The word preview matters more than a headline rate. Response speed depends on selected model, account tier, request length, capacity, network and the tools surrounding a request. A chat experience can feel faster without establishing a portable serving benchmark, and a provider-reported maximum does not predict the time needed to search, upload a document, execute code or wait for an external service.


## Search, files and code execution made Le Chat a collection of different workflows [source 1](https://mistral.ai/news/all-new-le-chat/) [source 2](https://mistral.ai/news/mistral-chat/)

The February post described access to recent information, image and document understanding, and a code interpreter for sandboxed execution, analysis, visualizations and simulations. Those are not interchangeable capabilities. A web-grounded answer depends on what sources were retrieved and how they were represented; a document answer depends on pages and structure that were actually processed; an executed calculation depends on code, inputs and the sandbox’s environment. Calling all three a “chat answer” can hide where a result came from.
Mistral’s earlier November record is useful context because it describes beta web search with citations, Canvas and document understanding around Pixtral Large. It also shows why historical feature records need date separation. The November beta framing does not prove the scope or maturity of the February product, while the February announcement does not rewrite the earlier model or partner information. A team comparing releases needs to preserve both the feature state and the date of the supporting record.


## Plan labels and private preview set different access boundaries [source 1](https://mistral.ai/news/all-new-le-chat/)

Mistral said most Le Chat features were available without charge and identified a Pro tier beginning at $14.99 a month, with Team and Enterprise options. This is a dated product statement, not current pricing guidance. Pricing, feature limits, regions and billing conditions can change, so an article about a February launch should not use the announcement as a later purchase recommendation. It does show that Mistral was attaching an explicit commercial structure to a product previously described in beta-oriented terms.
Le Chat Enterprise was described as a private preview for deployment in a customer environment with custom tools and customized models. A private preview is an invitation to evaluate a limited product, not general availability, an assurance of a specific security posture, or a statement that any tool may act on corporate systems. Connecting a chat assistant to documents, email, messaging or databases requires its own access rules and audit path; the post itself listed connectors and multi-step agents as forthcoming.


## A richer assistant interface did not make every output verified work [source 1](https://mistral.ai/news/all-new-le-chat/) [source 2](https://mistral.ai/news/mistral-chat/)

Le Chat’s release combined conversational response, generated images, uploaded files and executable analysis in one interface. That can shorten early research and drafting, but it also makes provenance easier to lose. A useful workflow should retain source links for search claims, the source file for document claims, the code and inputs for calculations, and a human review for an image or written assertion that will be published. The interface can coordinate those artifacts without proving their accuracy or authorization.
This record stays with February 6. It does not assign later research modes, connectors, memory controls or current service terms to the launch. Nor does it make health, finance, legal or operational decisions from an assistant response. The durable historical point is that Mistral broadened Le Chat from a conversational product into a multi-tool assistant with consumer and enterprise routes, while tool-specific validation and permissions remained outside the announcement.


## Sources
1. [Mistral AI: The all new Le Chat](https://mistral.ai/news/all-new-le-chat/)
2. [Mistral AI: Mistral has entered the chat](https://mistral.ai/news/mistral-chat/)

## Related reading
- [Amazon introduced Alexa+ as a generative-AI assistant with service experts and staged access](https://rangoon.ai/insights/archive-amazon-alexa-plus-february-2025/)
- [Mistral Code presented an enterprise coding stack rather than a single completion model](https://rangoon.ai/insights/archive-mistral-code-june-2025/)
- [Meta AI app launch made personalization and social discovery product-surface choices](https://rangoon.ai/insights/archive-meta-ai-standalone-app-2025/)

## More information

- [Documentation index](https://rangoon.ai/llms.txt)
- [AI access and policies](https://rangoon.ai/ai/)
- [Source repository](https://github.com/hypler-dev/rangoon)