Webapp, API and Playground Different Results

[herve76]

Am I the only one seeing huge differences between API and Playground results?

I am using the same payload as the Playground but I get very different results.

Any idea? or suggestions?

[PaulBellow]

You are not the only one!

[_j]

I can demonstrate basically getting the same results, yesterday:

Incomplete Output of 4o through API

[image] [image] First, you can press the ‘view code’ button to ensure you have the correct parameters. Then despite those, crank the max_tokens value you can receive as output way up. You can see my truncation at 128 tokens. Then look at the finish_reason in the response. If it is “stop”, the AI just decided to stop writing. Another oddity about this model is that specifying temperature and top_p values too low is not just a setting to zero, it a setting to “useless”. The parsing of negati…

You are sending to the same AI model. The only thing that might be different is the parameters you set, or the natural variation between responses when not constraining the outputs with temperature or top_p.

You are not the only one not using the API the same way as the playground sends.

[herve76]

I am using the same temperature and all the same parameters.

Note: IMG-2-TEXT

[herve76]

I saw them but nothing recent, and still no answer from OpenAI.