Working styles differ sharply on AutomationBench. GPT-5.5 (xhigh) takes a more action-intensive approach, averaging 49 tool calls across 25 turns per task, while Claude Opus 4.8 (max) is more…

Gemini 3.5 Flash just dropped. It’s the highest AutomationBench score yet. We benchmark every major model on real workflows across Sales, Marketing, Ops, Support, Finance, and HR, so you know what…

Yeah, I hope to run some open source models against the benchmark in the near future.