Mistral Next

Mistral Next Generative AI Model Quietly Starts Testing on Discord

Generative AI startup Mistral AI has quietly begun testing a new model currently named Mistral Next. Currently only available through the Direct Chat tab on the Large Model Systems Organization (LMSYS) page, Mistral Next was revealed only in a Discord comment from Mistral AI engineer Lélio Renard Lavaud and shared by NLP scientist Casper Hansen on X (formerly Twitter).

Mistral Next

There aren’t any details about Mistral Next’s parameters or training, but some comments by initial testers claim it is exceleelnt at logical reasoning and writing code. Some even say it’s at or above OpenAI’s GPT-4 model in that regard. Mistral may be quietly peforming well above its relative fame. The Mistral Medium model, part of its La Platforme hub of chat models, sits at number six on the LMSYS leader board, below only four versions of GPT-4 and Google’s Gemini Pro model.

The French generative AI startup has been rapidly accumulating capital and accolades.  Mistral emerged from stealth with a $113 million seed round last June, followed by a $415 million Series A funding round led by Andreessen Horowitz in December. And the company’s published LLMs are drawing plenty of interest. They include the Mistral 7B model that came out when the company debuted and the more recent Mixtral 8x7B LLM. Both are under an open Apache 2.0 license, aiming to demonstrate the capabilities of even modestly sized open-source LLMs for generative AI projects. Mistral Next would presumably also fall under that license. It’s not just as raw models that Mistral has drawn interest. Web browser Brave made Mixtral 8x7B the default model for its generative AI assistant Leo.

Brave Installs Mistral’s Mixtral as Default LLM for Generative AI Browser Assistant Leo

Mistral AI Raises $415M and Releases New LLM as Free Torrent

French Generative AI Startup Mistral Raises $113M