# The Same Agent Model Scores 62% in One Harness and 33% in Another

Hugging Face published a multi-harness reinforcement-learning recipe that trains through a traffic-capturing proxy, while a new arXiv study finds pooled cross-harness credit teaches configuration adaptation rather than transferable skill.

- Published: 2026-10-03T06:33:24.401Z
- Canonical: https://polylog.news/ai/2026-10-03/the-same-agent-model-scores-62-in-one-harness-and-33-in-anot
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [Hugging Face](https://huggingface.co/blog/sergiopaniego/trl-openenv-harness-training), [arXiv](https://arxiv.org/abs/2609.04518)

A small model scored 62% in the Mini-SWE-Agent harness and 33% in the Claude Code harness, with no training difference between the two runs. That 29-point spread, reported in a Russian-language writeup of Hugging Face's multi-harness guide,…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-03/the-same-agent-model-scores-62-in-one-harness-and-33-in-anot (subscription information: https://polylog.news/pricing).