HELM Robo-Reward-Bench - Robo Reward Bench Discrete: leaderboard
The "Robo Reward Bench Discrete" 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: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 (2025-04-14) | 1.4 | #250 |
| 2 | GPT-4o (2024-11-20) | 1.44 | #378 |
| 3 | Gemini 2.0 Flash | 1.66 | #344 |
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-robo-reward-bench-discrete · How It Works · Data refreshed daily, snapshot 2026-10-23.