Samuel Colvin

I think the API can be as Simple as ```py model = FallbackModel('openai:gpt-4o', 'claude-3-5-haiku-latest', GeminiModel(...)) agent = Agent(model, ...) ``` E.g. the signature of `FallbackModel` is something like ```py class...

Another advantage of having a consistent way to detect 503 etc. responses, is we can use it to mark live tests as xfail when the underlying model fails.

and should also be supported by #227 which will be released today.

@fatelei are you intending to work on this further?

@sydney-runkle please can you finish this off.

This would be great plase @pawamoy.

https://github.com/pydantic/pydantic-core/blob/a017bd9230561aab9868564045d3d37d30a3637a/build.rs#L16 You might have some luck setting the `PYTHON` env var., or looking at the PR that changed that code. My guess is the poetry is supposed to set the...

You shouldn't need to install typing-extensions first, it's a build requirement https://github.com/pydantic/pydantic-core/blob/a017bd9230561aab9868564045d3d37d30a3637a/pyproject.toml#L4

Yup, like the link I sent above, we use the default `python3` executable, or the value of the `PYTHON` env var to run the script that builds the core schema...

It sounds like the issue is with `pyo3_build_config::get().executable` then. Related to #674. @davidhewitt is on pat leave right now, but might have some thoughts when he's back. You could perhaps...