STORY · VERKTOY_
Prime Intellect launches verifiers v1 for more efficient agent training
Prime Intellect has launched verifiers v1, a redesigned environment for agentic reinforcement learning that reduces computational complexity from O(n²) to O(n) by storing rollout-traces as message DAGs. This enables practical testing of long agent tasks, as demonstrated by a 100B model running 40-turn SWE agent tasks on 6 H200 nodes in under 2 days.
WHY IT MATTERS
This represents a significant shift in how coding agents are trained and evaluated by making long-horizon tasks computationally feasible. At the same time, we're seeing a paradigm shift where "harnesses" are becoming central product surfaces for agent development, and cost per task is becoming the new benchmarking metric instead of token price.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.