Open Source · Mistral · Nvidia · Llama · AI Inference · Hugging Face
Mistral published Mistral Large 4 (Le Chonk), a 1-trillion-parameter multimodal model
Compiled by KHAO Editorial — aggregated from 2 sources. See llms.txt for citation guidance.
✓ KHAO Verified
Mistral Large 4 (aka Le Chonk) is a multimodal, multilingual, Mixture-of-Experts (MoE) model from Mistral.
Key facts
- Mistral Large 4’s text decoder has 72 layers, a hidden dimension of 6,144, and 64 attention heads
- On coding, ML4 achieves 61.7% on DeepSWE v1.1, 59.4% on SWE-Atlas-QnA, and 28.3% on Terminal-Bench 4
- Mistral Large 4 is also available on Inference Endpoints, with a ready-to-deploy recipe on 8 x NVIDIA H200 GPUs
- The 2-bit quants run on a single Apple Silicon Ultra computer with 512 GB of unified RAM
Summary
There are also GGUFs created by Unsloth and the llama.cpp team. Mistral Large 4 is particularly strong for cybersecurity, coding agents, and multimodality. "Welcome Le Chonk: Mistral Large 4 on Hugging Face" Table of Contents What is new with Mistral Large 4 Cybersecurity Capabilities Architecture Vision encoder Benchmark Results Inference Support Transformers vLLM Llama.cpp Fine-tuning with TRL Inference Endpoints Dell Enterprise Hub Try it out. Mistral Large 4 combines a large pool of experts with sparse attention and a Pixtral vision encoder.