# Mistral Large 4 "Le Chonk": 1T Parameters, 49B Active, Open Weights Due October 27

Mistral's new model, Mistral Large 4, features about 1 trillion parameters and is currently in public preview. Open weights are expected by October 27…

**Published:** 2026-10-07 · **Author:** Yash Thakker
**Source:** https://explainx.ai/newsletter/mistral-large-4-le-chonk-1t-parameters-49b-active-open

---
> **In short:** Mistral's new model, Mistral Large 4, features about 1 trillion parameters and is currently in public preview. Open weights are expected by October 27, 2026, making it a significant development in multimodal AI capabilities.

## Mistral Large 4 "Le Chonk": 1T Parameters, 49B Active, Open Weights Due October

### Mistral Large 4 "Le Chonk": 1T Parameters, 49B Active, Open Weights Due October 27

Mistral has launched a public preview of its new model, Mistral Large 4, featuring about 1 trillion parameters and 49 billion active ones. Open weights are promised by October 27, 2026. The model is designed for multimodal tasks, handling both text and images.

**What you can do:** Consider exploring the Mistral API for early access to test its capabilities before the open weights are released.

[Source](https://explainx.ai/blog/mistral-large-4-le-chonk-1t-open-weights-preview-2026)

## AI-Driven Cyberattacks on South Korean Banks

### South Korea Bank Hacks: What AI Actually Did, and What Builders Should Learn

South Korea is investigating a suspected AI-assisted cyberattack that targeted seven financial firms, exposing data of about 66,000 individuals. While AI tools are suspected, the specific technology used remains unconfirmed. Emergency security checks are underway across financial institutions.

[Source](https://explainx.ai/blog/south-korea-ai-bank-hacks-artex-credential-stuffing-october-2026)

### OpenAI just dropped 700 preprints of mathematical proofs and counterexamples

OpenAI has released 700 preprints containing mathematical proofs and counterexamples. This collection aims to contribute to the field of mathematics, providing new insights and discussions for researchers and enthusiasts alike.

[Source](https://github.com/openai/math/tree/main/preprints)

### An Open Letter to Steven Pinker on AI

An open letter addressed to Steven Pinker discusses various perspectives on artificial intelligence. It raises important questions about the implications of AI technology and its impact on society, inviting thoughtful dialogue on the subject.

[Source](https://www.astralcodexten.com/p/an-open-letter-to-steven-pinker-on)

## Understanding Recent Claims on 3SUM and APSP Hypotheses

### Did Claude Just Break the 3SUM Conjecture? What Is Claimed and What Is Verified

Recent reports suggest that an algorithm discovered by Anthropic's Claude may refute significant computational hypotheses, including the 3SUM conjecture. However, independent verification of these claims is still pending, meaning caution is advised before accepting them as established facts.

**What you can do:** Stay informed about the developments surrounding this algorithm. If you're involved in computational research or applications, consider how these hypotheses impact your work. Engage with the community to discuss the implications of these claims as more information becomes available.

[Source](https://explainx.ai/blog/claude-3sum-apsp-claim-algorithm-lean-what-is-verified-2026)

## Explore AI Safety in Our Free Workshop

### AI Safety & Best Practices — Free Live Hour

Join our upcoming workshop on AI safety concepts essential for builders and operators. Starts on October 24, 2026.

[Source](https://explainx.ai/workshops/ai-safety)

## Reader poll

What do you think about the potential of Mistral Large 4 as a multimodal AI model?

- Very promising
- Somewhat promising
- Not very promising
- Uncertain/Not applicable

## Sources

- [Mistral Large 4 "Le Chonk": 1T Parameters, 49B Active, Open Weights Due October 27](https://explainx.ai/blog/mistral-large-4-le-chonk-1t-open-weights-preview-2026)
- [South Korea Bank Hacks: What AI Actually Did, and What Builders Should Learn](https://explainx.ai/blog/south-korea-ai-bank-hacks-artex-credential-stuffing-october-2026)
- [OpenAI just dropped 700 preprints of mathematical proofs and counterexamples](https://github.com/openai/math/tree/main/preprints)
- [An Open Letter to Steven Pinker on AI](https://www.astralcodexten.com/p/an-open-letter-to-steven-pinker-on)
- [Did Claude Just Break the 3SUM Conjecture? What Is Claimed and What Is Verified](https://explainx.ai/blog/claude-3sum-apsp-claim-algorithm-lean-what-is-verified-2026)
- [AI Safety & Best Practices — Free Live Hour](https://explainx.ai/workshops/ai-safety)

---

From [explainx.ai](https://explainx.ai/newsletter). Read online, comment and subscribe: https://explainx.ai/newsletter/mistral-large-4-le-chonk-1t-parameters-49b-active-open