https://promptfoo.dev/ logo
execution timeout parameter
# ❓|questions
z
I’m seeing an issue that some LLM call hang forever. For example, I’m testing both gpt-4o and gemini. gpt-4o returned while gemini hang there. Is there any parameter that controls the timeout of LLM calls?
i
hmm, by default there is a 5 min timeout that Gemini should respect. Is that not working?
You can override it with the
REQUEST_TIMEOUT_MS
environment variable, e.g.
REQUEST_TIMEOUT_MS=10000 promptfoo eval
z
Thanks