The web search tool is currently limited to GPT-4o and GPT-4.1 models, which have a maximum output of 32,768 tokens despite a large 1,047,576-token context window. This output token limit feels a bit restrictive for web search models.
Given the size of the context window, is there any plan to increase the max output token limit for these web search models?
That’s a fair ask. Was there a particular report you were trying to generate, or a use case you have in mind that would leverage a higher max output token limit?
For what it’s worth, the o3 and o4-mini models with 100,000 max output tokens are industry leading at the moment (in terms of output token limit), and I might say equally suitable for web search with their 200,000 context windows.