CompanionBench - English - Selfobject Responsiveness: leaderboard
Metric: Mean judge score (1-10). Source: arxiv.org. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 8.38 |
| 2 | Claude Opus 4.8 | 8.35 |
| 3 | DeepSeek V4 Pro | 8.26 |
| 4 | GPT-5.4 | 8.23 |
| 5 | Claude Sonnet 4.6 | 8.14 |
| 6 | GPT-5.1 | 8.14 |
| 7 | Qwen 3.7 Max | 8.12 |
| 8 | GLM-5 | 8.11 |
| 9 | Qwen 3 235B A22B 2507 Instruct | 8.09 |
| 10 | Kimi K2.6 | 8.08 |
| 11 | GLM-5.1 | 8.04 |
| 12 | DeepSeek V4 Flash | 8.02 |
| 13 | Qwen 3.5 397B A17B | 8.01 |
| 14 | GLM-5.2 | 8 |
| 15 | Kimi K2.5 | 7.99 |
Interactive version: theaggregate.ai/benchmark?slug=companionbench-english-selfobject-responsiveness · How It Works · Data refreshed daily, snapshot 2026-09-19.