Anyone finding success with LLM’s other than 4o? I feel like it is the only one that can do tool calls but latency still seems high