It’s Wrestlemania season for WWE fans but for us AI fans, it’s Modelmania
Just in the last 24 hours, we’ve got updates on three of the largest LLMs:
1/ GPT-4 Turbo is out of preview - GPT-4 Turbo (with Vision) is now generally available in
@OpenAI’s API. Vision requests can now also use JSON mode and function calling. OpenAI says that this update also improved the model in maths and reasoning.
2/ Gemini 1.5 Pro Now Available.
@Google’s model with a 1M token context window is now available in Gemini’s developer API. It also accepts audio files—an upgrade from the limited version in AI Studio.
3/
@MistralAI dropped another mysterious torrent link. Looks like a 176B MoE model. From the Discord chatter we’ve gone through, it feels like a model better than Sonnet but not as good as GPT-4.