O3-mini Completion API Access Incredibly Slow?

Trying to figure out what’s going on here… I’ve been using o1 in the API at medium reasoning and it’s been acceptable speed-wise.

I’m now dropping in o3-mini to test it out and when accessed via the API it’s so slow it’s unusable (network timeout slow). The same conversation in the Playground seems about on-par with o1 (expected faster but at least it works). We’re not using streaming in this case.

I’m not sure where to start here since the conversation between o1 and o3-mini is the same.

I’m in Tier 5 in case that matters. Input tokens 7,906 and output 1,992.

Well, I may have spoken too soon and this was just a temporary glitch because things seem more or less back to where I’d expect them at this point. Good times!

2 Likes

Phew! Let us know if you run into it again.

3 Likes

I have the same issue. Assistant via API or playground response is very slow or even fails “The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center”.
My current model 4o-mini works without any problems.
Could you please check?

1 Like

I was able to reproduce the issue in the playground using the Assistants API with o3-mini low and send a message to OpenAI.

Hope this helps!

1 Like

Hey team, thanks for flagging this. The team is currently aware of the latency issues with Assistants API for o3-mini and we’re currently investigating it. We’ll update back once its resolved!

2 Likes