I need some help. I’m using the API Responses endpoint, but I’m still relying on GPT-4o, because GPT-5 (even on low-effort settings) feels extremely slow.
The issue is most noticeable when using the web search tool:
On GPT-5, the first response with a web search takes over 1 minute.
On GPT-4o, the same request returns results in just 5–10 seconds.
Is this expected behavior for GPT-5 right now, or could there be an issue with my setup?
Welcome to the community, @JakaSpaka.
gpt-5 is a reasoning model by default, while gpt-4o is just a chat completion model. Thus, it normally takes a bit longer than 4o to respond with the final answer.
This is not “a bit longer”. A bit longer would mean 5sec vs 15sec. Not So long that users thing that chat doestn work, even tho i give back INFO that is working on it. It ranges from 1 min +. Even i tried all effort minimum, as found on API.
We use gpt-5-mini with the web search tool and the response API. Our responses take no where near one minute. Things you can try:
“tools”: [{“type”: “web_search”, “search_context_size”: “high”, “user_location”: {“type”: “approximate”, “country”: “US”}}],
“reasoning”: {“effort”: “low”},
“text”: {“verbosity”: “low”},
“service_tier”: “priority”,
or could there be an issue with my setup?
What exactly is your setup?