Replace retired Claude model ids in the Anthropic examples and README

[kasparovabi]

📥 Pull Request

📘 Description

The three Anthropic examples and the three Anthropic snippets in the README call model ids that Anthropic has retired, so a reader who runs them as written gets an API error instead of a trace. This swaps each id for the current model in the same tier and changes nothing else. 15 lines across 7 files, every one of them a model= argument.

Where	Was	Now	Retired on
examples/anthropic/agentops-anthropic-understanding-tools.py and .ipynb (4 calls each)	claude-3-7-sonnet-20250219	claude-sonnet-5	19 February 2026
examples/anthropic/anthropic-example-sync.py and .ipynb (1 call each)	claude-3-7-sonnet-20250219	claude-sonnet-5	19 February 2026
examples/anthropic/anthropic-example-async.py and .ipynb (1 call each)	claude-3-7-sonnet-20250219	claude-sonnet-5	19 February 2026
README.md Anthropic sync, streaming and async snippets (3 calls)	claude-3-opus-20240229	claude-opus-5	5 January 2026

Dates are from https://platform.claude.com/docs/en/about-claude/model-deprecations. The notebooks and the .py files were changed together so they keep matching. None of the calls pass temperature, top_p or top_k, which matters because Opus 5 rejects non-default values for those, so nothing else needed to move.

Left alone on purpose. docs/v2/examples/anthropic.mdx carries the same ids but is generated from the notebook by add-notebook-examples-to-docs.yml on push to main, so editing it here would only create a conflict with that workflow. llms.txt is likewise generated by compile-llms-txt.yml. The tests/ fixtures, VCR recordings and test_attributes.py reference old ids as recorded data and still work as tests. docs/v2/integrations/anthropic.mdx uses claude-sonnet-4-20250514 and claude-opus-4-20250514, which are also past retirement, but that is hand-written documentation rather than a runnable example and I did not want to widen this diff. Happy to send that separately if you want it.

🧪 Testing

examples-integration-test.yml runs these three example files on pull requests, so CI will exercise the change against the API with your key. I have not run them against the API myself. I did check that all three notebooks still parse as JSON after the edit and that git diff shows only the model= lines.

I read this diff line by line before opening it, and I am happy to adjust or drop any part of it. What brought me here was a checker I maintain that flags retired Claude model ids, so I am saying that up front rather than leaving you to wonder. The claim that the old ids no longer work rests on the published retirement dates, I have not reproduced the failure myself. If you would rather handle this your own way, or would rather not get this kind of contribution at all, tell me and I will not send another.