Product Thumbnail

Mistral Large

Cheaper GPT-4 rival that supports 32K-token context windows

Open Source
Developer Tools
Artificial Intelligence
GitHub

Mistral Large has top-tier reasoning capacities, is multi-lingual by design, has native function calling capacities and a 32k model. The pre-trained model has 81.2% accuracy on MMLU.

Top comment

Comment highlights

Mistral AI put France in the center of the open source AI movement. This is amazing to see Mistral on PH!

Saying it's cheaper in a pay-to-go plan service isn't entirely true. Maybe for users who don't use it much, it's cheaper, but for those of us who use it daily with a large token output, I don't think it is, especially knowing that the cost is $8/Million input tokens and $24/Million output tokens. I think it's very risky to say that without falling into a lie for some users. Beyond that, I believe it's an extremely solid product. It seems to have a great future. This type of competition is what improves the industry. Keep it up!

Well, I just have to say that it looks like a new game-changer in the industry. Big congrats on the launch, and best of luck today, team ❤️🦄

Congratulations on your launch guys! You just did an amazing tool! So impressive.

Very cool analog, keep it up, good luck with the implementation of the platform guys, I think it will surely beat the top analogs