SignalFrontier releaseSG-0017
Mistral previewed Mistral Large 4, an open-weight model it ranks among the top five at cybersecurity
Mistral released a public preview of Mistral Large 4 and said it would publish the weights by the end of the month. Mistral says the model ranks among the top five on the Artificial Analysis Cyber Index and scores 82% on a test that asks models to reproduce and patch a real vulnerability, a task on which it says Claude Opus 5.5 and GPT-6 Astra score near zero because they refuse. Before the weights release, cybersecurity leaders, vetted partners and state authorities are testing a version with reduced moderation and expanded cyber capabilities.
Why it matters
Open weights let anyone run, modify or remove the safeguards of a model with near-top cyber skills, outside any provider's monitoring. Mistral presents the lack of refusals on vulnerability work as an advantage for defenders, while reporting higher refusal rates than other open models on malicious cyber prompts. The benchmark figures are the developer's own and do not establish real-world exploit capability.
Sources
Read the reporting
1 source. Links go to the original publishers; the summary above is in our own words.
Signals
More from the news desk
Cite and share
Use this signal
Citation
Paperclip Index. “Mistral previewed Mistral Large 4, an open-weight model it ranks among the top five at cybersecurity.” Signal SG-0017. Reported Oct. 6, 2026; updated Oct. 7, 2026. https://paperclipindex.com/signal/SG-0017