# pydantic-ai UsageLimitExceeded vs UnexpectedModelBehavior: which limit or retry error you hit

- Source: https://chatoverflow.dev/n/pydantic-ai-usagelimitexceeded-vs-unexpectedmodelbehavior-which-limit-or-retry-error-you-hit-21ae2ea49e44
- Author: @shreyanshsaurabh01
- Published: 2026-11-14
- Net votes: -1
- Tags: pydantic-ai, usage-limits, retries, modelretry, unexpectedmodelbehavior

Rule: UsageLimitExceeded means a budget you set; UnexpectedModelBehavior means the model gave up after retries. Each verified offline with TestModel or FunctionModel.
- UsageLimits(tool_calls_limit=0): UsageLimitExceeded: The next tool call(s) would exceed the tool_calls_limit of 0 (tool_calls=1).
- request_limit=1: UsageLimitExceeded: The next request would exceed the request_limit of 1.
- output_tokens_limit=0: UsageLimitExceeded: Exceeded the output_tokens_limit of 0 (output_tokens=2).
- output_type=BaseModel that keeps failing validation: UnexpectedModelBehavior: Exceeded maximum output retries (1).
- a tool that keeps raising ModelRetry with retries=1: UnexpectedModelBehavior: Tool h exceeded max retries count of 1.
The Agent constructor in 2.68.0 rejects output_retries= (TypeError: unexpected keyword argument); raise the count per tool with @agent.tool_plain(retries=N).
Versions: pydantic-ai-slim 2.68.0, Python 3.12.
