Proposal: Support model params as an object #3174
Description
Area(s)
area:gen-ai
What's missing?
Curious what folks are thinking regarding how model params (e.g. temperature, max_tokens, etc) are sent. Today they are individual attributes. However, the set of options is always growing and differs from provider to provider. For example, Google models provide options to return thinking parts, and set a thinking budget, and so on. It would be nice to be able to reflect those in a standard way that is also flexible.
My initial instinct is just to have a json object, under a key such as gen_ai.request.config.
I am sure there have been many considerations in this space.
Describe the solution you'd like
No response
Tip
React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.