Mistral introduces Le Chonk: Mistral Large 4 with a trillion parameters
Mistral Large 4 is available in preview with a trillion total parameters, 49 billion active parameters, and training on 3800 Grace Blackwell accelerators. API pricing is high, while weights are promised by the end of the month.
Mistral released Le Chaton Fat Le Chonk
Mistral Large 4 is a trillion-parameter model with 49 billion active parameters, trained on 3800 Grace Blackwell accelerators in Mistral's own datacenter. On most benchmarks it falls slightly short of the best open models, but at least it is in the same league.
The model is already available in preview, with API pricing at $1.36/$4.18 per million tokens, which is fairly expensive for such a model. Weights are promised at the end of the month. RL is still ongoing, so the model may improve further.
Why it matters
AnalysisIt shows Mistral is trying to compete with the strongest open-weight models, but high API prices and delayed weights limit immediate accessibility.
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.
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.
Alibaba launches Qwen4-Max with 1M-token context
Alibaba’s new flagship model is positioned for coding, game development and agent tasks, with a 1 million token context window and lower pricing than top US competitors.
Part I. AI Agent Architecture: The Foundations
The eleven foundational building blocks of modern autonomous agents: model taxonomy, the agentic runtime loop, advanced prompting, distillation, skills architecture, memory systems, function calling, and FinOps.