Mistral releases ML4 with open weights – coding results rise to 41 percent
Mistral AI has released the open weights for Mistral Large 4. The model achieves 41 percent on Terminal-Bench and consumes 47 percent fewer tokens per task.
What happened?
On 30 October 2026, French company Mistral AI released the open model weights for Mistral Large 4 (ML4), internally codenamed 'Le Chonk'. The model features approximately one trillion parameters and offers three reasoning modes: low, medium, and high. The result on the coding benchmark Terminal-Bench 4.0 increased to 41 percent, compared with 28 percent in the preview version.
Key facts
| Lanseringsdatum | 30 oktober 2026 |
|---|---|
| Terminal-Bench 4.0-poäng | 41 % (upp från 28 %) |
| Minskad tokenförbrukning | 47 % färre tokens per uppgift |
| Hårdvara vid träning | 4 000 Nvidia Grace Blackwell GPUs |
Why it matters
The release demonstrates that open models are approaching the performance levels of proprietary systems in agent-based coding. According to the Artificial Analysis Coding Agent Index, ML4 reached a score of 48, placing it close to competitors such as GLM-5.3 and Codex GPT-6.1 Terra, while the model consumes on average 47 percent fewer tokens per task.
Who is affected?
The news primarily concerns software developers, AI engineers, and technology companies looking to run advanced coding agents locally or in their own data centres. Organisations with strict data sovereignty requirements within the EU also gain a competitive open alternative to closed APIs.
Impact on the EU
As a European company with data centres within the EU, Mistral provides businesses with a local alternative for data processing. The model complies with EU requirements and offers organisations within the union access to open weights without reliance on US cloud infrastructure.
What else you should know
ML4 was first introduced in a preview on 6 October 2026, before the open weights were published following three weeks of partner testing. Training was conducted from scratch over two months on 4,000 Nvidia Grace Blackwell GPUs.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Är modellen tillgänglig inom EU?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "Mistral releases ML4 with open weights – coding results rise"