Mistral — the French AI company that built its reputation on efficient, open-weight models — launched its 128 billion parameter flagship model on May 3, 2026. It ships with two notable features: an async cloud coding environment and a new "Work" agentic mode in Le Chat (Mistral's chat interface). This isn't the scrappy open-source challenger anymore — it's a direct competitor to Claude and GPT for professional use cases.
Key Takeaway
Mistral's 128B model is the first European frontier model that genuinely competes with US labs on capability. The Work agentic mode turns Le Chat from a simple chatbot into a task-executing agent — similar to what Microsoft is doing with Agent 365 but in a consumer-accessible product.
What Is Work Agentic Mode?
Work mode in Le Chat lets Mistral's model execute multi-step tasks autonomously — research, planning, tool use, and output generation in sequence without manual prompting at each step. This is the same "agentic" pattern that Karpathy described at Sequoia AI Ascent: AI that doesn't just respond to queries but plans and executes workflows.
The practical difference: in standard chat mode, you ask a question and get an answer. In Work mode, you describe a goal ("research the top 5 competitors in X market and create a comparison table with pricing, features, and market share") and the agent handles the research, analysis, and formatting autonomously.
How Does the 128B Model Compare?
| Model | Parameters | Agentic Mode | Async Coding | Open Weights |
|---|---|---|---|---|
| Mistral 128B | 128B | Yes (Work mode) | Yes (Le Chat) | Partial (some variants) |
| Claude Opus 4.7 | Undisclosed | Via Claude Code | No | No |
| GPT-5.4 | Undisclosed | Via Codex | Yes (Codex) | No |
| Gemini 3.1 Ultra | Undisclosed | Limited | No | No |
For a comprehensive comparison of all AI coding tools available right now, see our best AI coding tools ranking. And to get better results from any AI model, try the free Prompt Optimizer.
---📬 Getting value from this? We track every major model launch. Get it in your inbox →
---Frequently Asked Questions
Is Mistral's 128B model better than Claude or GPT?
On raw benchmarks, it's competitive but not leading — Claude Opus 4.7 and GPT-5.4 still outperform on most evaluations. Mistral's advantage is the integrated Work mode and competitive pricing, especially for European companies with data sovereignty requirements.
Can I use Mistral's model via API?
Yes. Mistral offers API access alongside Le Chat. The 128B model is available for direct API integration. Pricing is competitive with OpenAI and Anthropic.
Is Mistral open source?
Partially. Mistral has released open-weight versions of some models but the 128B flagship has more restrictive licensing. Check Mistral's current licensing terms for your use case.
Disclosure: Some links in this article are affiliate links. We only recommend tools we've personally tested and use regularly. See our full disclosure policy.