Mistral Releases Open Weights for Large 4, the 1-Trillion-Parameter Le Chonk Model
Mistral Releases Open Weights for Large 4, the 1-Trillion-Parameter Le Chonk Model
Mistral has released open weights for its flagship Large 4 model, internally called Le Chonk, and reduced API pricing.
Mistral has introduced its new flagship model, Large 4. Internally, the company officially calls it Le Chonk — a reference to the Le Chaton Fat meme, the fake giant Mistral model that kept the community’s attention for months.
The open weights for Mistral Large 4 are available on Hugging Face: Mistral-Large-4-1T-A52B.
Three weeks after the announcement, the model received additional training that noticeably improved results in coding and cybersecurity tasks.
API pricing has also been lowered to $0.70 per million input tokens and $2.10 per million output tokens.
Why it matters
AnalysisOpen weights and lower API prices make a trillion-parameter model more accessible for builders, researchers, and security-focused applications.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Mistral AI (European Open and Efficient Models)
A series of high-performance open and commercial models from the French startup Mistral AI (Mistral 7B, Mixtral 8x7B, Mistral Large). Known for their compactness, speed, and respect for data privacy.
Mistral Publishes Le Chonk Weights as Polymarket and Binance Hacks Are Reported
Online chatter says Mistral has released Le Chonk weights, with reports that Polymarket and then Binance were compromised shortly afterward.
LLM (Large Language Model)
A fundamental class of neural network architectures based on autoregressive transformers, predicting the probabilistic distribution of subsequent tokens and demonstrating emergent properties of abstract reasoning, code synthesis, and logical inference.
Part II. AI Agent Architecture: From Model to Memory and Tools
A deep engineering dive: tokenization, KV-cache, sampling determinism, the training stack, quantization, context engineering, the safety stack, production RAG, ReAct, and multi-agent swarm systems.