Latency Issues
Hey everyone,
I’ve noticed that the latency during the agent step has become quite high — it’s taking around 20 seconds to process through the knowledge base and generate a response.
I’m using GPT-5 Mini with moderate reasoning and 5000 max tokens. It was running much faster before, but lately the response time has dropped significantly.
Not sure if it’s a temporary performance bug or something on my end, but wanted to flag it in case others are seeing the same slowdown.
3 Replies
not GPT-5 Mini, but i tried GPT-5 for a while with moderately high reasoning and 3500 tokens. it very often timed out, so i went back to GPT 4.1
Try different models. FYI, it might be due to OpenAI, you might check their status.
agreed, i def prefer Claude (mainly Sonnet 4) over OpenAI's LLMs for agent steps! (i had been trying to figure out which of OpenAI's LLMs would work best for one particular agent step, but it's still not as good as Sonnet 4)