Post

Wrote up my initial impressions of the new Google Gemini 2.0 Flash model - it's really good, and the streaming mode (where you can stream video and audio to it and get audio streamed right back) is pure science-fiction simonwillison.net/2024/Dec/11/...

simonwillison.netGemini 2.0 Flash: An outstanding multi-modal LLM with a sci-fi streaming modeHuge announcment from Google this morning: Introducing Gemini 2.0: our new AI model for the agentic era. There’s a ton of stuff in there (including updates on Project Astra and …

Carlos Lemus

@carloslemus.bsky.social

12 Dec 2024

Hi, the code with the requests library seems to have executed fine when I tried it using LibreChat.

Simon Willison

@simonwillison.net

12 Dec 2024

That's weird, their documentation says "Code cannot access the network" www.librechat.ai/docs/feature...

librechat.aiCode Interpreter APIExecute code securely and manage files seamlessly with LibreChat's Code Interpreter API

Carlos Lemus

@carloslemus.bsky.social

12 Dec 2024

Worked again for Bluesky. It's either using knowledge of titles from its dataset, or it actually does have network access 🤔

Simon Willison

@simonwillison.net

That looks to me like it's using its knowledge - does that tool provide any kind of UI indication when it's actually executing code? It should do: the ChatGPT and Claude UIs both have clear code execution visual hints, which is definitely the right design decision

12/12/24, 2:28 pm0 reposts0 likes

Carlos Lemus

@carloslemus.bsky.social

12 Dec 2024

No, I made the false assumption based on your blog that the Flash 2.0 backend itself executed some code. It does not. But it can infer what the output *would* be for a given piece of code. TIL