HELM Robo-Reward-Bench - Bridge: leaderboard
Metric: Absolute Error. Source: crfm.stanford.edu. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (xHigh) | 0.45 |
| 2 | GPT-5 Mini (2025-08-07) | 0.53 |
| 3 | Gemini 3 Pro (Preview) | 0.75 |
| 4 | GPT-5 | 0.79 |
| 5 | GPT-5.2 | 0.94 |
| 6 | Gemini 3 Flash (Preview) | 0.99 |
| 7 | Gemini 2.5 Flash Lite | 1.07 |
| 8 | Gemini 2.5 Flash | 1.08 |
| 9 | Llama 4 Maverick Instruct | 1.13 |
| 10 | Gemini 2.5 Pro | 1.17 |
| 11 | GPT-5 Nano | 1.31 |
| 12 | Llama 4 Scout Instruct | 1.44 |
Interactive version: theaggregate.ai/benchmark?slug=helm-robo-reward-bench-bridge · How It Works · Data refreshed daily, snapshot 2026-09-08.