# Mistral AI introduces Mistral Large 4 in public preview [Canonical reading](https://escapemeridian.com/readings/mistral-ai-introduces-mistral-large-4-in-public-preview-73b624dc) Mistral AI’s post, dated October 6, 2026, announces a public preview of Mistral Large 4, a model the company describes as open-weight and multimodal. Mistral says the model performs strongly across cybersecurity, coding, agentic workflows, science, and knowledge work, citing results from benchmarks and human evaluations. The company highlights a cybersecurity test in which it says the model reproduced and patched a real software vulnerability, and argues that open weights and self-deployment can let organizations run security work under their own policies. These are Mistral’s claims about evaluations and potential use, not evidence here of performance in independent real-world deployments. Mistral says it is red-teaming the preview with cybersecurity leaders, vetted partners, and state authorities, and plans to release the weights by the end of October. It also says the model’s reinforcement learning run is still in progress and forecasts further improvement as training scales. The post does not provide the additional architecture, benchmark, and post-training details it says will follow. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source Mistral AI · Article Original source date: Oct 6, 2026 (2026-10-06T12:00:27.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-10-31T15:08:56.768Z. First observed: 2026-10-06T13:23:12.172Z. Retrieved: 2026-10-24T04:37:37.349Z. - [Introducing Mistral Large 4](https://mistral.ai/news/mistral-large-4/) ## Limits Text as retrieved on October 24, 2026; originally published October 6, 2026. The post presents Mistral’s own claims about the model and its evaluations; the supplied evidence does not establish independent real-world performance. The post says further architecture, benchmark, and post-training details will be shared, but they are not included in this reading’s evidence. ## Open question How will Mistral Large 4 perform in independent real-world security work, and what will the additional evaluation details show? ## Source-stated Mistral AI announced a public preview of Mistral Large 4 on October 6, 2026. > October 6, 2026 > By Mistral > Share this post - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says Mistral Large 4 is an open-weight multimodal model and reports benchmark and evaluation results across several areas. > ML4 is a 1 trillion-parameter natively multimodal model with 52 billion active parameters. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says the model scored 82% on an Artificial Analysis Cyber Index test involving reproducing and patching a real vulnerability. > ML4 scores 82%, the highest of any model. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says it plans to release the model weights by the end of October. > We will release the weights by the end of the month. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says the reinforcement learning run is ongoing and forecasts large and rapid improvements as training scales. > As we scale up training on our expanded infrastructure, we expect large and rapid improvements in the weeks and months to come. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source and contributor context Mistral AI · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Open weights A model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted. - [Definition reference](https://opensource.org/ai/open-weights) ## Multimodal model A model that works with more than one kind of input or output, such as text, images and audio. - [Definition reference](https://developers.google.com/machine-learning/glossary#multimodal-model) ## Agentic AI AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer. - [Definition reference](https://developers.google.com/machine-learning/glossary#agentic) ## Benchmark A defined set of tests used to compare systems. Results depend on the tasks and measures selected. - [Definition reference](https://arxiv.org/abs/2211.09110) ## Red-teaming Deliberately attacking or tricking a system to find its weaknesses before others do. - [Definition reference](https://en.wikipedia.org/wiki/Red_team) ## Reinforcement learning Training by trial and error, in which a system is rewarded for the outcomes its designers want. - [Definition reference](https://developers.google.com/machine-learning/glossary#reinforcement-learning-rl) ## Mistral AI Mistral AI develops models and products with several options for how organizations deploy them. - [Explore this context](/entities/mistral-ai) - [Reference](https://mistral.ai/news/) ## Editorial provenance Version 1. Publication ID: 4dfa6f4c-8b3c-b4bb-b89d-ff4958790d9a. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-31T15:08:56.620Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: 0f96b2b5b1cd39961172acae42f2b97b0a5372d3968252bb6285dcaed34d5c22. Source text hash: 44fff5e357977e03d358c0711971c39fcf1956024374a8b5635df2c83698349d. ## Revision and source-check history ### Version 1 — 2026-10-31T15:08:56.768Z Mistral AI’s post, dated October 6, 2026, announces a public preview of Mistral Large 4, a model the company describes as open-weight and multimodal. Mistral says the model performs strongly across cybersecurity, coding, agentic workflows, science, and knowledge work, citing results from benchmarks and human evaluations. The company highlights a cybersecurity test in which it says the model reproduced and patched a real software vulnerability, and argues that open weights and self-deployment can let organizations run security work under their own policies. These are Mistral’s claims about evaluations and potential use, not evidence here of performance in independent real-world deployments. Mistral says it is red-teaming the preview with cybersecurity leaders, vetted partners, and state authorities, and plans to release the weights by the end of October. It also says the model’s reinforcement learning run is still in progress and forecasts further improvement as training scales. The post does not provide the additional architecture, benchmark, and post-training details it says will follow. ## Original source Mistral AI · Article Original source date: Oct 6, 2026 (2026-10-06T12:00:27.000Z). Date basis: Date supplied by the publisher’s feed. Added to Escape Meridian: 2026-10-31T15:08:56.768Z. First observed: 2026-10-06T13:23:12.172Z. Retrieved: 2026-10-24T04:37:37.349Z. - [Introducing Mistral Large 4](https://mistral.ai/news/mistral-large-4/) ## Limits Text as retrieved on October 24, 2026; originally published October 6, 2026. The post presents Mistral’s own claims about the model and its evaluations; the supplied evidence does not establish independent real-world performance. The post says further architecture, benchmark, and post-training details will be shared, but they are not included in this reading’s evidence. ## Open question How will Mistral Large 4 perform in independent real-world security work, and what will the additional evaluation details show? ## Source-stated Mistral AI announced a public preview of Mistral Large 4 on October 6, 2026. > October 6, 2026 > By Mistral > Share this post - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says Mistral Large 4 is an open-weight multimodal model and reports benchmark and evaluation results across several areas. > ML4 is a 1 trillion-parameter natively multimodal model with 52 billion active parameters. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says the model scored 82% on an Artificial Analysis Cyber Index test involving reproducing and patching a real vulnerability. > ML4 scores 82%, the highest of any model. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says it plans to release the model weights by the end of October. > We will release the weights by the end of the month. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source-stated Mistral says the reinforcement learning run is ongoing and forecasts large and rapid improvements as training scales. > As we scale up training on our expanded infrastructure, we expect large and rapid improvements in the weeks and months to come. - [Supporting source](https://mistral.ai/news/mistral-large-4/) ## Source and contributor context Mistral AI · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Open weights A model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted. - [Definition reference](https://opensource.org/ai/open-weights) ## Multimodal model A model that works with more than one kind of input or output, such as text, images and audio. - [Definition reference](https://developers.google.com/machine-learning/glossary#multimodal-model) ## Agentic AI AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer. - [Definition reference](https://developers.google.com/machine-learning/glossary#agentic) ## Benchmark A defined set of tests used to compare systems. Results depend on the tasks and measures selected. - [Definition reference](https://arxiv.org/abs/2211.09110) ## Red-teaming Deliberately attacking or tricking a system to find its weaknesses before others do. - [Definition reference](https://en.wikipedia.org/wiki/Red_team) ## Reinforcement learning Training by trial and error, in which a system is rewarded for the outcomes its designers want. - [Definition reference](https://developers.google.com/machine-learning/glossary#reinforcement-learning-rl) ## Mistral AI Mistral AI develops models and products with several options for how organizations deploy them. - [Explore this context](/entities/mistral-ai) - [Reference](https://mistral.ai/news/) ## Editorial provenance Version 1. Publication ID: 4dfa6f4c-8b3c-b4bb-b89d-ff4958790d9a. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-10-31T15:08:56.620Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: 0f96b2b5b1cd39961172acae42f2b97b0a5372d3968252bb6285dcaed34d5c22. Source text hash: 44fff5e357977e03d358c0711971c39fcf1956024374a8b5635df2c83698349d. ### passed — 2026-10-31T15:08:56.620Z — anthropic/claude-opus-5.5 Publication: 4dfa6f4c-8b3c-b4bb-b89d-ff4958790d9a attribution — pass: This is a single Mistral AI blog post. The reading attributes the performance, safety and roadmap statements to Mistral throughout, using phrasing such as 'Mistral says' and 'the company describes'. It does not present the company's claims as established fact. > Today, we’re launching a public preview of Mistral Large 4. summary — pass: The reading leads with the actor and the development: a public preview of an open-weight multimodal model. It keeps the post's central cybersecurity argument, that open weights and self-deployment let organizations run security work under their own policies. It also notes the planned weight release, the ongoing reinforcement learning run and the details still to come. Central terms are defined beside the reading. The source's point that closed models refuse the vulnerability task could be stated more directly, but its omission does not prevent understanding. > giving organizations both the capability and the autonomy to run advanced security work under their own policies date — pass: The reading's October 6, 2026 date matches originalPublished (2026-10-06T12:00:27Z) and the post's dateline, at day precision. 'End of the month' in an October post is fairly rendered as end of October. The improvement forecasts are attributed to Mistral rather than presented as outcomes, and there is no hindsight framing. A later dateModified (Oct 7) exists, but no editedAfterPublication flag is set. > October 6, 2026 By Mistral support — pass: Each claim matches a source passage: the 1 trillion-parameter multimodal description, the 82% score on the vulnerability reproduce-and-patch test, the weight-release plan, and the ongoing RL run with forecast improvements. The caveat that these are Mistral's own claims, not independently demonstrated real-world performance, is appropriate. The question follows from the evidence. > We will release the weights by the end of the month, along with more details on the architecture, additional benchmarks, and our post-training methodology. media — pass: No verified media candidates were supplied, and the reading selects none. This is consistent with the evidence. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/73b624dce5711f73ce6fab076b590ea40495e8b1299da9f1052111a3a38241de/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/mistral-ai-introduces-mistral-large-4-in-public-preview-73b624dc.json)