▶️ Watch on YouTube: https://youtu.be/CABVrfh2ALo
🌐 Website: https://tekthrill.com
📺 Subscribe: https://youtube.com/@tekthrill
Europe's Mistral AI just released the weights of Mistral Large 4, its 1-trillion-parameter model nicknamed "Le Chonk". Here is what was actually released, what changed since the October 6 preview, how it really compares with rival open and closed models, the cybersecurity debate, and what it means for businesses, including in India.
Chapters:
0:00 Intro: Mistral Large 4 goes open
0:25 What was released
1:04 What changed since the preview
1:52 How it really compares
2:35 The cybersecurity debate
3:29 What this means for you
4:32 The bigger picture
Key facts (with sources):
• Released as open weights on October 29, 2026, after a public preview on October 6 (Mistral AI).
• About 1.05 trillion total parameters, 52 billion active per token, mixture-of-experts design, text and image input, 1 million token context (Mistral docs).
• Trained from scratch on 3,800 Nvidia Grace Blackwell GPUs in Europe (Mistral, via The Next Web).
• Terminal-Bench 4.0: 28% at preview, 41% at release; 48% on the Artificial Analysis Coding Agent Index, tied with GLM-5.3 while using 47% fewer tokens (Mistral's figures).
• In Surge's blind tests it was preferred over GLM-5.3, Kimi K3 and DeepSeek V4.1 Flash, but won only 22% of comparisons against Claude Opus 5.5 (Mistral, citing Surge).
• Fifth-best open-weights model on the Artificial Analysis Intelligence Index as of October 29 (Hugging Face).
• 51% on the Artificial Analysis Cyber Index and a 96.6% refusal rate on malicious cyber requests, per Mistral, which also says it sees potential for uplift to conduct real attacks and is modifying its guardrails.
• A quantised version runs on four Nvidia B200 or B300 GPUs; smaller distilled versions are planned with no dates (The Next Web).
• License: Mistral Ecosystem Open-Weight License, which Hugging Face describes as an MIT derivative with limits on model-as-a-service offerings.
Benchmarks quoted here are mostly published by Mistral itself; treat them as company claims.
Would you run an open trillion-parameter model in your company, or stick with closed AI? Tell us in the comments 👇
Try free AI tools 👉 https://tekthrill.com
Subscribe to TEKTHRILL for daily AI and tech news! 🚀
Sources:
Mistral AI (open-weights release): https://mistral.ai/news/mistral-large-4-oss/
Mistral AI (preview announcement): https://mistral.ai/news/mistral-large-4/
Mistral Docs (model card): https://docs.mistral.ai/models/mistral-large-4-0
The Next Web: https://thenextweb.com/news/mistral-large-4-open-weights-release
Hugging Face blog: https://huggingface.co/blog/mistral-large4
Mistral AI on X (teaser video): https://x.com/MistralAI/status/2107457414387622310
