AARRI-Bench (Hermes Agent) - Hands-on: leaderboard

Metric: Task success rate (%) under the 0/1 reward on the hands-on tasks: coding, experimental setup and data processing; Harbor containerized runs, Hermes Agent agent harness; a configuration's tasks without a valid run are left out of its denominator; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 6 models tracked.

Top models

#ModelScore
1Qwen 3.6 Plus64.3
2Claude Opus 4.757.1
3MiniMax-M2.757.1
4Claude Sonnet 4.650
5DeepSeek V4 Flash50
6Qwen 3 235B A22B 2507 (Thinking)38.5

Interactive version: theaggregate.ai/benchmark?slug=aarri-bench-hermes-agent-hands-on · How It Works · Data refreshed daily, snapshot 2026-09-29.