I have provided o1 with a list of tools and the expectation is that o1 figures out a list of tools to call and returns all of them in a single response (like other gpt models). However, what I’ve observed is that o1 returns only a single tool call at a time, this increases the latency quite a bit if there are several tool calls that need to be called.
Has anyone else observed a similar issue? Curious to know how you’ve solved it.