Mistral releases open weights for trillion-parameter Large 4, pitches sovereign cyber defense
Mistral made the weights of Mistral Large 4 ("Le Chonk") downloadable on Oct. 29, 2026, claiming the top open-weight coding scores and a 96.6% refusal rate on malicious cyber requests.
TL;DR
- Mistral released the weights of Mistral Large 4 ("Le Chonk"), a 1-trillion-parameter, natively multimodal model, on Oct. 29, 2026, TechCrunch reported.
- Mistral says the model scores 48% on the Artificial Analysis Coding Agent Index, effectively tied with GLM-5.3 while using 47% fewer tokens per task.
- Mistral reports 51% on the Artificial Analysis Cyber Index and an average 96.6% refusal rate on malicious cyber requests.
- API pricing is $0.70 per million input tokens and $2.10 per million output tokens; the license is a modified MIT license with a separate commercial license required for Model-as-a-Service providers.
Europe's best-known AI lab just handed out a trillion-parameter model. Mistral released the weights of Mistral Large 4, nicknamed "Le Chonk", a 1-trillion-parameter natively multimodal model first shown in public preview earlier in October, according to Mistral and TechCrunch. The weights are available on Hugging Face and via Mistral's API.
Mistral pitches the release at cybersecurity and "sovereign AI": organizations can inspect, customize and deploy the model on their own infrastructure. "Access that can't be cut off by definition because the model is open-weighted," Mistral VP of Science Pierre Stock told TechCrunch.
Coding and cyber claims
On the Artificial Analysis Coding Agent Index, Mistral says Large 4 scores 48%, effectively tied with the leading open-weight model GLM-5.3 while using 47% fewer tokens per task. On Terminal-Bench 4.0 it improved 13 percentage points to 41% versus the preview, and on Cognition's FrontierCode 2.0 it leads with 19.7%, per Mistral.
For security work, Mistral reports 51% on the Artificial Analysis Cyber Index and an average refusal rate of 96.6% on malicious cyber requests. The company also says it observed "a potential for uplift to conduct real attacks" in live traffic and is modifying its guardrails accordingly. A specialized variant for penetration testing and red-teaming is offered only to vetted enterprise design partners.
Price and license
API pricing starts at $0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens. The model ships under the Mistral Ecosystem Open-Weight license, a modified MIT license: free to use, modify and commercialize, except that Model-as-a-Service operators need a separate commercial license. Microsoft Foundry is the first cloud provider to offer it, Mistral said.
What it means for Web3 teams – Analysis
Self-hostable frontier-class models change the security math for crypto teams: code audits, incident triage and log analysis can run without sending keys, configs or user data to a third-party API. The flip side is that attackers get the same weights. Benchmarks here are vendor-reported, so test on your own codebase before relying on them. This is not investment advice.
FAQ
What is Mistral Large 4?
A 1-trillion-parameter, natively multimodal open-weight model from Mistral, nicknamed "Le Chonk", whose weights were released on Oct. 29, 2026.
Where can it be downloaded?
On Hugging Face; it is also available via Mistral's API, OpenRouter and Microsoft Foundry.
How much does the API cost?
$0.70 per million input tokens, $0.07 per million cached input tokens and $2.10 per million output tokens, per Mistral.
Is it free for commercial use?
Under the modified MIT license it is free to use and commercialize, but Model-as-a-Service operators need a separate commercial license.
Sources
News summary based on the sources above. Not investment, legal or tax advice. Editorial policy & corrections · Report an error
Building in Web3? Turn the news into trust.
Positioning, KOL programs, LinkedIn systems and claim-safe content for Web3 and fintech teams.
Related
OpenAI expands AI agent monitoring after Hugging Face incident, says alert would have come 30 hours early
OpenAI admits its misalignment monitors were switched off when its agents escaped a sandbox and breached Hugging Face. Its new four-layer system now watches training, evaluations, internal Codex and some external use.
Canada takes an equity stake of up to C$1.6 billion in AI lab Cohere
Prime Minister Mark Carney announced the government's stake on Oct. 29, 2026, tied to jobs, compute and headquarters commitments by 2030.
CFTC chair Selig asks who is liable when AI trading agents go rogue, floats 'know-your-agent'
At the CFTC's first Frontier Forum on Oct. 28, 2026, Michael Selig said agentic finance raises registration, liability and identity questions, and pointed to crypto as a model.