HELM Robo-Reward-Bench - Utokyo Xarm Bimanual Converted Externally To Rlds: leaderboard
The "Utokyo Xarm Bimanual Converted Externally To Rlds" subtask of HELM Robo-Reward-Bench. Stanford HELM's Robo-Reward-Bench: vision-language models act as reward models over robot manipulation episodes, predicting task progress or success from video.
Metric: Absolute Error. Source: crfm.stanford.edu. Saturation forecast: Not forecast. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (2026-04-23) | 1.28 | #24 |
| 2 | GPT-5.5 (xHigh) | 1.35 | #29 (GPT-5.5) |
| 3 | GPT-5 | 1.37 | #104 |
| 4 | GPT-5.1 (2025-11-13) | 1.44 | #105 |
| 5 | Gemini 2.5 Flash Lite | 1.46 | #421 |
| 6 | Llama 4 Maverick Instruct | 1.46 | #460 |
| 7 | GPT-5 Nano | 1.52 | #428 |
| 8 | GPT-5 Mini (2025-08-07) | 1.59 | #174 |
| 9 | Llama 4 Scout Instruct | 1.61 | #542 |
| 10 | Gemini 3 Flash (Preview) | 1.65 | #84 |
| 11 | Gemini 2.5 Flash | 1.68 | #242 |
| 12 | GPT-5.2 | 1.72 | #113 |
| 13 | Gemini 2.5 Pro | 1.75 | #153 |
| 14 | Gemini 3 Pro (Preview) | 1.79 | #66 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-robo-reward-bench-utokyo-xarm-bimanual-converted-externally-to-rlds · How It Works · Data refreshed daily, snapshot 2026-10-23.