HelpBench: leaderboard
Metric: Overall rubric score (%): points for met positive criteria and avoided negative criteria (factual criteria written per question by privacy, safety and security experts, plus shared delivery criteria) over the maximum points, applied by a Gemini 2.5 Pro auto-rater at temperature 0; chat versions with default parameters, mean of five responses per question, 450 questions; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.3 | 87 |
| 2 | GPT-5 Chat | 87 |
| 3 | GPT-4.1 | 84 |
| 4 | Claude Opus 4.6 | 84 |
| 5 | Gemini 3 Flash | 84 |
| 6 | Claude Sonnet 4.6 | 83 |
| 7 | Grok 4 | 83 |
| 8 | Qwen 3.6 Plus | 83 |
| 9 | Gemini 3.1 Pro (Preview) | 82 |
| 10 | Claude Sonnet 4 | 82 |
| 11 | GLM-5.1 | 82 |
| 12 | Gemini 2.5 Pro | 81 |
| 13 | GLM-4.6 | 80 |
| 14 | Grok 4.20 | 80 |
| 15 | DeepSeek V3.2 (Non-reasoning) | 80 |
Interactive version: theaggregate.ai/benchmark?slug=helpbench · How It Works · Data refreshed daily, snapshot 2026-09-29.