Mistral AI just released its most powerful model yet. The benchmarks are pretty wild. Their new 124B parameter Pixtral-Large multimodal model.
- Outperforms top models on math reasoning and real-world tasks
- Beats Gemini 1.5 Pro and GPT-4o on chart and document understanding
- Features a 128k context window, handling a min. of 30 high-res images
Mistral’s Le Chat platform also got revamped, complete with web search, document analysis, and image generation (powered by Black Forest Labs’ Flux Pro).

Six months ago, the gap between open and closed models was significant.

Now, state-of-the-art AI (with all the latest features) is quickly becoming available to anyone who wants to use it.
Try on HuggingFace: https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411
Read related articles: