Post

Wrote up my initial impressions of the brand new Claude Sonnet 4.5 - I think it may live up to Anthropic's claims of being the "best coding model in the world", for the next few weeks at least! simonwillison.net/2025/Sep/29/...

simonwillison.netClaude Sonnet 4.5 is probably the “best coding model in the world” (at least for now)Anthropic released Claude Sonnet 4.5 today, with a very bold set of claims: Claude Sonnet 4.5 is the best coding model in the world. It’s the strongest model for building …

29/9/25, 6:13 pm26 reposts158 likes

Bite Code!

@bitecode.dev

29 Sept 2025

We need this t-shirt, but with "Evolution of LLM" and Pelican test results

Jochen Voss

@seehuhn.bsky.social

30 Sept 2025

Do you have any sense of whether completely replaces the more expensive Opus 4.1? Or are there still use cases where Opus 4.1 is better?

Simon Willison

@simonwillison.net

30 Sept 2025

It feels like a total replacement to me, like how Sonnet 3.5 totally outperformed Opus 3.

Simon Willison

@simonwillison.net

29 Sept 2025

Where it really shines is in their new claude.ai Code Interpreter mode - I had it checkout my GitHub repo, install dependencies, run tests and experiment with a complex new feature, all prompted from the web browser on my iPhone simonwillison.net/2025/Sep/29/...

Not Juha Lankinen

@jlankinen.bsky.social

29 Sept 2025

I’m wondering if there’s somewhere a reference pelican on a bike to compare to.

Simon Willison

@simonwillison.net

29 Sept 2025

I don't think there is, because pelicans can't ride bicycles!

Juanmi

@juanmirod.bsky.social

29 Sept 2025

I came to bluesky just to see the pelican that Claude Sonnet 4.5 had made 😂 Thanks for your analysis! the pelican is really good! I am starting to think they train them on this task :P have you tried other animals or other drawings to see if the results are wildly different?