Harness-Zero distills agent harnesses into base models
Harness distillation shifts the value of hand-built agent scaffolding from runtime glue into permanent model capability, which could change how labs decide what to ship as harness versus as weights. Teams building agent frameworks can use the corrected-run pipeline to convert an optimized harness into training data for their own models. Compared with the code-as-harness alternative, the agent-as-harness form is the stronger option for frontier models here.
A Google-affiliated paper (arXiv 2609.24974) introduces Harness-Zero, a method for distilling an optimized agent harness into a model.
Removing the specialized harness raises macro task success from 23.3% to 44.
3%, exceeding the 41.7% the base model achieves with the harness attached.
During training, an agent guided by the optimized harness corrects student responses in the deployment action space, and these corrected runs become training demonstrations.
Across 28 harness-induced behaviors in knowledge work, tool use and science, 82.3% are recovered on average.
For frontier models on the same evolved harness, the agent-as-harness form also beats the code-as-harness form.
Super interesting paper from Google and colleagues. It studies where it's possible to distill an agent harness. With the specialized harness removed, macro task success goes from 23.3% to 44.3%. That is higher than the 41.7% the base model reaches with the harness attached. Harness-Zero uses the optimized harness only during training. The optimized harness and the deployment harness have different action spaces, so a harnessing agent guided by the optimized harness corrects the student's responses in the deployment action space before they run. Those corrected runs become the training demonstrations. Across 28 harness-induced behaviors in knowledge work, tool use and science, 82.3% are recovered on average. For frontier models using the same evolved harness, the agent-as-harness form also beats the code-as-harness form. It remains to be seen how robust the approach is, but it's very interesting to see potential in harness distillation. Paper: arxiv.org/abs/2609.24974 Chat with Paper: academy.dair.ai/papers/harness…📊 10 ⚡
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Harness-Zero, Google, agent harness distillation, agents.
Source attribution: curated from elvis (@omarsar0).
