Mistral moves coding agents off the laptop

Mistral released Medium 3.5, a 128B dense model with a 256k context window, in public preview under a modified MIT license. It becomes the default model in Le Chat and replaces Devstral 2 in the Vibe CLI. Pricing is $1.5 per million input tokens and $7.5 per million output tokens. Reasoning effort is configurable per request. Mistral reports 77.6% on SWE-Bench Verified and 91.4 on τ³-Telecom, and says self-hosting is possible on as few as four GPUs. The bigger shift is where the agents run. "Coding agents have mostly lived on your laptop. Today we're moving them to the cloud, where they run on their own, in parallel, and notify you when they're done," Mistral writes. Vibe remote agents can be launched from the CLI or from Le Chat, run in isolated sandboxes, and open a pull request on GitHub when finished. Local CLI sessions can be teleported to the cloud with history and approvals intact. Vibe integrates with GitHub, Linear, Jira, Sentry, Slack, and Teams. A new Work mode in Le Chat runs multi-step tasks — inbox triage, meeting prep, research briefs, Jira ticket creation — calling tools in parallel until the job is done.