Mistral Large 4 Launches on Hugging Face With Open Weights and Text-and-Image Input

2 reports 2 reporting sources 35 minutes agoNew

Understand the event

AI overview

Mistral Large 4 is now available on Hugging Face with open weights and launch-day support for inference and fine-tuning, under the Mistral Ecosystem Open-Weight License, which limits model-as-a-service offerings. Also called Le Chonk, the multilingual mixture-of-experts model accepts text and images and generates text. According to Hugging Face, the model has about 1T total parameters, a 1M-token context window, and FP8 and NVFP4 variants. Hugging Face also said the model was trained in Mistral’s own data centers and described it as excelling at cybersecurity.

Generated by AI from reports · Updated 13 minutes ago

Latest development10/29 23:30 (UTC+8)

Hugging Face interviews Mistral’s Pierre Stock and Sagar Vaze on Mistral Large 4 Hugging Face · X

Models in this story

Mistral Large 4

API prices of all models

Developments

2 developments

10/29 23:30 (UTC+8) · 1 reports
Hugging Face Shares a Conversation with Mistral on Mistral Large 4
Hugging Face · X：Hugging Face interviews Mistral’s Pierre Stock and Sagar Vaze on Mistral Large 4
10/29 08:00 (UTC+8) · 1 reports
Hugging Face Introduces Mistral Large 4 and Its Capabilities
Hugging Face Blog：Mistral Large 4 arrives on Hugging Face with multimodal capabilities and open weights

Report timeline

Follow the reports to explore different sides of the event.

10/29

23:30
Hugging Face · X
Hugging Face interviews Mistral’s Pierre Stock and Sagar Vaze on Mistral Large 4
Hugging Face sat down with Mistral’s Pierre Stock and Sagar Vaze on launch day of Mistral Large 4. Open weights. Trained in Mistral’s own data centers. Described as excelling at cyber. Full conversation on cyber, training, evals, and sandboxing: https://t.co/JOyEpqGBO9
08:00
Hugging Face BlogPicks
Mistral Large 4 arrives on Hugging Face with multimodal capabilities and open weights
Mistral Large 4, also called Le Chonk, is a multilingual MoE model that accepts text and images and generates text, with about 1T total parameters and a 1M-token context window. Hugging Face details its FP8 and NVFP4 variants, cybersecurity benchmarks, and launch-day inference and fine-tuning support. The weights use the Mistral Ecosystem Open-Weight License, which limits model-as-a-service offerings.

Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue

Event heat trend

There is not enough continuous observation data to plot a trend yet.