llm-anthropic 0.30
Simon Willison’s Weblog Simon Willison
Simon Willison shipped llm-anthropic 0.30 with Claude Sonnet 5.5 support. It can now refresh Anthropic models and count tokens before you send a prompt.
Based on reporting by Simon Willison’s Weblog, Simon Willison — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Simon Willison has released llm-anthropic 0.30, the plugin that gives llm access to Anthropic models, including the Claude series. The update adds support for Claude Sonnet 5.5 and two new commands that make the whole thing a bit less static and a bit more useful.
The bigger change is llm anthropic refresh. Instead of waiting for a new release every time Anthropic adds a model, the command can pull the current model list straight from Anthropic’s API. That removes a small but annoying bit of maintenance work. It also means support can move at the speed of Anthropic’s own catalog, not the release cadence of the plugin.
There’s also llm anthropic count, which taps Anthropic’s free token counting API. Before a prompt goes out, it can tell you how many tokens it will use. That sounds modest, but it’s exactly the kind of feature people end up wishing they had after a few expensive or oversized prompts.
The release was posted on 28 September 2026 at 11:06 pm. Small update, practical payoff. No drama, just fewer reasons to babysit model lists and token guesses by hand.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI tooling: less ceremony, more plumbing. The best closed-model integrations aren’t the ones that shout the loudest; they’re the ones that quietly stop becoming stale the moment a vendor moves. A refresh command and a token counter won’t win any keynote slides, which is probably why they’ll actually get used.
Read more about this at: Simon Willison’s Weblog
