Mistral Releases Weights for Trillion-Parameter Mistral Large 4 Model on Hugging Face
The Facts
- Mistral released the weights for Mistral Large 4 (ML4), a trillion-parameter open-source AI model nicknamed 'Le Chonk', which was first introduced earlier in October. TechCrunch only
- The weights are available on Hugging Face and via Mistral's API, allowing businesses and governments to download and deploy the model themselves. TechCrunch only
- Mistral says ML4 performs comparably to China's open-weight GLM-5.3, particularly on cyber tasks, while using 47 percent fewer tokens per task. TechCrunch only
- Mistral claims ML4 outranks all other open models on Artificial Analysis' Coding Agent Index for agentic coding tasks. TechCrunch only
- Mistral is positioning ML4 for cyber defense, saying enterprises and public-sector institutions can deploy it to detect, analyze and respond to threats across code and IT infrastructure. TechCrunch only
Dive deeper · 1 more factShow less
- Pierre Stock, Mistral's VP of Science, said open weights mean access to the model cannot be cut off and that it can be deployed on-premises. TechCrunch only
Context
French AI company Mistral has released the weights of Mistral Large 4, a trillion-parameter model, for download and self-deployment. The company is marketing the model for cybersecurity use by enterprises and governments that want control over their data and systems.
Performance and efficiency claims, including the comparison with GLM-5.3 and the coding index ranking, come from Mistral and were not independently verified in the source. Mistral CEO Arthur Mensch said the company aims to deliver state-of-the-art performance at a fraction of the compute and capital used by closed labs.
How the Agent Panel Scored This
Independent AI personas each scored importance 1–10; the story's score is a trimmed mean (highest and lowest votes dropped). Vote spread: 1.
| 5 | News Editor | A concrete release of open weights for a trillion-parameter model is a real industry event with some reach for enterprises and governments, but it is a single company's product launch with unverified performance claims and limited broad public impact, so it sits in the moderate range. |
| 5 | Policy Expert | A concrete release of open weights for a major trillion-parameter model is a notable industry event with some downstream deployment impact, but it is a product release from a single company with vendor-only, unverified claims and limited immediate effect on many people. |
| 5 | Science & Tech Analyst | Release of open weights for a trillion-parameter frontier-class model is a notable concrete AI industry event with some enterprise and government relevance, but the performance claims are unverified, it is from a single source, and its impact is largely confined to the AI and cybersecurity community. |
| 5 | Business Analyst | A major open-weight trillion-parameter frontier model release is notable for the AI industry and cybersecurity market, but it is a product launch with unverified vendor claims and no immediate broad economic impact. |
| 5 | Critical Reader | A concrete release of open weights for a trillion-parameter model is a real event with some downstream deployment impact, but the performance claims are unverified, the model was previously introduced, and it is a single-source vendor announcement of moderate importance. |
| 4 | Informed Citizen | A major open-weight model release from a notable AI company is a concrete event of moderate significance for tech and enterprise users, but its direct impact on ordinary people is limited and the performance claims are unverified. |
| 4 | Global Affairs Analyst | A notable open-weight frontier AI model release with cyber-defense positioning is a concrete industry event with moderate geopolitical relevance, but its impact is limited to the AI sector and its claims are unverified. |