429 (Requests rate limit exceeded) status code from Mistral API are not handled anywhere

[TypeHintsFun]

Initial Checks

• I confirm that I'm using the latest version of Pydantic AI
• I confirm that I searched for my issue in https://github.com/pydantic/pydantic-ai/issues before opening this issue

Description

429 (Requests rate limit exceeded) status code from Mistral API are not handled anywhere, what makes it quite problematic to use it

pydantic_ai just throws arror like this:
pydantic_ai.exceptions.ModelHTTPError: status_code: 429, model_name: mistral-medium-latest, body: {"message":"Requests rate limit exceeded"}

Python, Pydantic AI & LLM client version

python 3.12.8, pydantic_ai 0.2.12, mistralai 1.81

[TypeHintsFun]

I just checked and found out that in pydantic_ai there is no system of trying to repeat action after errors like Rate Limit. I think, we should add it

[TypeHintsFun]

Also, it will be nice to add rps, rpm, rph and other limitations like this into UsageLimits (or into some new object, maybe RateLimits)

[DouweM]

@TypeHintsFun Agent supports a retries argument. Can you try increasing that from the default of 1 and see if it works?

Proper rate limiting support is also being implemented in #1734.

[TypeHintsFun]

@DouweM, I just checked it. No, the retries argument does not definitely affect 429 processing from Mistral

[TypeHintsFun]

Proper rate limiting support is also being implemented in github.com/#1734.

Hope it will approved soon

[DouweM]

@TypeHintsFun I'll be reviewing it today!