API response (Using gpt-5-mini) very slow now

API Bugs
1

I’m using the Responses API with the gpt-5-mini model. I’m on Priority mode but responses are extremely slow (over 5 minutes).
Current config:

response = await client.responses.parse(
    input=input,
    instructions=instructions,
    previous_response_id=previous_response_id,
    tools=tools,
    text_format=text_format,
    reasoning={"effort": "low"},
    text={"verbosity": "low"},
    model=model,
    parallel_tool_calls=True,
    tool_choice="auto",
    truncation="auto",
    timeout=60,
    service_tier="priority"
)

Please help—yesterday everything was normal.

1 Like
2

Hi @Bui_Thanh_Tinh,

I’d recommend checking the service_tier field in the Responses objects you received with higher latencies. This would indicate which service tier that particular request was processed under.

2 Likes
3

You are getting rate limited. I have a test suite I built that will do a few hundred API calls. Some stall out like this, then continue like normal. Depending on your use case, you can use a timer and resend the request. I helped, but did not fully fix, my issues by just funding my account to get to the next tier. Also, I noticed that the wording is a little ambiguous on the usage tier. You do not need to “use” the funds, just fund your account to get to the next tier.

1 Like