Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseAPI / Product · ActionableThe Decoder→ original

Mistral releases Vibe desktop app with parallel agent tasks

Mistral has released a desktop app for its Vibe agent (formerly Le Chat). You can hand it several tasks at once and watch it work on them in parallel. It can run in a local repository or in a cloud sandbox, and Git worktrees keep parallel tasks in the same project from interfering with each other. Computer control, like Claude's, is not available yet. For teams, it is handy for making several changes at once, but Mistral's models still trail the leaders.

AI-processed from The Decoder; edited by Hamidun News
Mistral releases Vibe desktop app with parallel agent tasks
Source: The Decoder. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceThe Decoder ↗
Editor-in-Chief
Key metric #1Available for macOS and Windows; a Linux version is planned
Key metric #2Vibe previously lived on the web, in the command line and on smartphones
Key metric #3Anthropic's Claude has been able to control apps and the browser since spring

France's Mistral has added a full desktop app to its Vibe agent, formerly known as Le Chat. The idea will be familiar to anyone who uses AI assistants for coding: instead of a single conversation where you wait for an answer to each task, you give the agent several assignments at once and watch it work on them simultaneously. This is not a breakthrough. It is Mistral bringing its product side up to par with competitors. Still, it is useful for developers and product managers because it shows what a standard agentic toolkit looks like in late 2026.

What users actually get

Until now, Vibe came in three forms: a web app, a command-line tool and a smartphone app. The desktop version combines the local capabilities of Vibe CLI with the cloud sessions from Vibe Code Web. In practice, the same agent can work in two modes: directly in your local repository, or in a cloud sandbox isolated from your computer.

The most practical detail here is Git worktrees. Imagine asking the agent to fix an authentication bug, update dependencies and write tests for a payment module at the same time. If all three tasks modify the same working folder, they will start stepping on each other's toes. Worktrees create a separate working copy of the same repository, with its own branch, for each task. The agents do not conflict, and you can review and merge each result separately. It is the same technique developers have long used by hand when they need to keep several branches open at once.

Beyond code, Work mode now has access to local files. That means you can use the agent for documents and folders on your computer, not just for a project. However, Mistral does not mention any ability to control the computer: opening programs, clicking through interfaces, filling in forms.

Where Mistral is still catching up

Here the gap with competitors is clear. Anthropic's desktop app has let Claude open apps, control the browser and fill in spreadsheets since spring. Vibe Desktop is so far limited to code and files. If you need an agent that can navigate service interfaces on its own, without an API, Mistral has no such offering right now.

The picture is similar with models. In October, Mistral closed some of the gap with the release of Large 4, but in head-to-head comparisons it still falls noticeably behind Gemini 4 Carbon and Claude Opus 5.5. It also trails the strongest open models from China. This matters for an agent: the quality of planning and coding depends largely on the model under the hood, and a polished shell does not make up for it.

Why it matters in practice

Mistral's business is not built on consumer apps. The company earns most of its revenue from enterprises and government bodies that run its models through an API or on their own servers. For them, a desktop agent is more of an extra item on the platform's feature list than the main product. European companies with data-residency requirements and a wish for independence from American vendors may find this package appealing precisely because it can be deployed on their own infrastructure.

What developers and product managers should do:

  • If you already work with Mistral through the API, Vibe Desktop is worth trying as a convenient interface for parallel tasks in a repository. Check the results of each branch separately: parallelism speeds up the work, but it does not remove the need for review.
  • If you are choosing an agent for tasks involving interfaces and a browser, look at solutions that already offer computer control.
  • If you are building your own agent product, take note of the combination of a local mode, a cloud sandbox and worktrees. It is becoming a baseline user expectation.

As for limitations, the app is available for macOS and Windows, and the Linux version is only planned. That is a nuisance for the developers who work on Linux. The source contains no independent tests of Vibe Desktop's quality, so it is too early to judge how well the agent handles real-world tasks.

Story

Key Questions Answered

What users actually get?

Until now, Vibe came in three forms: a web app, a command-line tool and a smartphone app. The desktop version combines the local capabilities of Vibe CLI with the cloud sessions from Vibe Code Web. In practice, the same agent can work in two modes: directly in your local repository, or in a cloud sandbox isolated from your computer. The most practical detail here is Git worktrees. Imagine asking the agent to fix an authentication bug, update dependencies and write tests for a payment module at the same time.

Where Mistral is still catching up?

Here the gap with competitors is clear. Anthropic's desktop app has let Claude open apps, control the browser and fill in spreadsheets since spring. Vibe Desktop is so far limited to code and files. If you need an agent that can navigate service interfaces on its own, without an API, Mistral has no such offering right now. The picture is similar with models.

Why it matters in practice?

Mistral's business is not built on consumer apps. The company earns most of its revenue from enterprises and government bodies that run its models through an API or on their own servers. For them, a desktop agent is more of an extra item on the platform's feature list than the main product. European companies with data-residency requirements and a wish for independence from American vendors may find this package appealing precisely because it can be deployed on their own infrastructure.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…