Wrong output token limit (2048, but should be 4096)

edit

The XDA article is simply wrong. The actual output token limit for GPT-4o in the API is 4,096 tokens as I've verified just now. It's the same for all of the previous GPT-4 and 3.5 models except GPT-4-32k. I don't know how to "verify" this with Wikipedia rules as technically it'll be first-hand knowledge (not from some "independent" source), but that's just a real fact. Maybe https://openrouter.ai/models/openai/gpt-4o can be good as a source? PURWANTO (talk) 13:39, 7 July 2024 (UTC)Reply