@omarsar0
Why it wins Most extra solves over OWL come from collaborative refinement enabled by A2A (52%), with smaller gains from reduced context redundancy (8%). How agents collaborate is key to these strong results. OWL’s few wins over Anemoi largely stem from worker stochasticity and web-agent latency.