Benchmarks · agentic
Hermes Bench
Uses mean reward across 150 Hermes Bench tasks in the Hermes Agent harness. Score: mean task reward. Collected by Jake Boggs' benchmark dashboard.
16 models · 16 results · data as of 2026-10-29 · portal.nousresearch.com ↗
Leaderboard
Opus 5.5 max
All results
16 rows| # | Model | Effort | Hermes Bench |
|---|---|---|---|
| 1 | Opus 5.5 | max | 76.7% |
| 2 | Grok 4.7 | max | 74.4% |
| 3 | GPT-6.1 Sol | max | 73.9% |
| 4 | GPT-6 Astra | max | 73.7% |
| 5 | GPT-6 Sol | max | 73.5% |
| 6 | Sonnet 5.5 | max | 73.5% |
| 7 | GLM 5.3 | max | 71.8% |
| 8 | Kimi K3 | max | 71.5% |
| 9 | Haiku 5.5 | max | 71.0% |
| 10 | DeepSeek V4.1 Flash | max | 70.7% |
| 11 | Qwen 3.8 Max | max | 70.5% |
| 12 | GPT-6 Luna | max | 70.5% |
| 13 | GLM 5.3 Flash | max | 68.9% |
| 14 | Gemini 3.8 Flash | max | 62.2% |
| 15 | Hy4 | high | 58.6% |
| 16 | Ling 3.0 Flash | 53.8% |