Mistral launches Large 4, tying GLM-5.3 at 48% on the Coding Agent Index

Top News · 1,291 HN points

Mistral positions Large 4, also called Le Chonk, as its flagship open-weight model. In Surge Code's blind coding evaluations, it ranked behind only Claude Opus 5.5, ahead of GLM-5.3, Kimi K3, and DeepSeek V4.1 Flash among the evaluated models.

Here's what changed:

• Artificial Analysis' Coding Agent Index gives it 48%, tied with GLM-5.3.
• Artificial Analysis separately reports 47% fewer tokens per task on average.
• API pricing is $0.70 per million input tokens, $0.07 per million cached input tokens, and $2.10 per million output tokens.
• Terminal-Bench 4.0 reaches 41%, up 13 percentage points from preview.
• Low, Medium, and High reasoning modes let you trade performance against cost. High is the default.

One catch: The Mistral Ecosystem Open-Weight license allows use, modification, fine-tuning, and commercialization, but commercial Model-as-a-Service operators need a separate license.

Try it: Start in Mistral Studio or use the Mistral API. You can also download the weights from Hugging Face and self-host with vLLM or SGLang.

Sources: mistral.ai