WolfBench (Hermes): leaderboard
Metric: Mean Success Rate (%, 89 Terminal-Bench 2.0 tasks). Source: wolfbench.ai. Saturation forecast: Rough model projection: around 2026. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol (WolfBench Run: agent=hermes; version=v2026.7.7.2; timeout=3600s; thinking=max; providers=openai; source-model=GPT-5.6%20Sol) | 84.49 |
| 2 | gpt-6-astra (WolfBench Run: agent=hermes; version=v2026.8.31; timeout=3600s; thinking=max; providers=openai; source-model=gpt-6-astra) | 83.37 |
| 3 | GPT-5.6 Terra (WolfBench Run: agent=hermes; version=v2026.7.7.2; timeout=3600s; thinking=max; providers=openai; source-model=GPT-5.6%20Terra) | 80.9 |
| 4 | GPT-5.6 Luna (WolfBench Run: agent=hermes; version=v2026.7.7.2; timeout=3600s; thinking=max; providers=openai; source-model=GPT-5.6%20Luna) | 78.65 |
| 5 | GPT-5.5 (WolfBench Run: agent=hermes; version=v2026.4.23; timeout=3600s; thinking=xhigh; providers=openai; source-model=GPT-5.5) | 74.16 |
| 6 | GPT-5.5 (WolfBench Run: agent=hermes; version=v2026.4.23; timeout=3600s; thinking=medium; providers=openai; source-model=GPT-5.5) | 72.81 |
| 7 | Gemini 3.5 Flash (WolfBench Run: agent=hermes; version=v2026.5.16; timeout=3600s; thinking=high; providers=google; source-model=Gemini%203.5%20Flash) | 70.34 |
| 8 | GLM-5.2 [W&B] (WolfBench Run: agent=hermes; version=v2026.6.19; timeout=3600s; thinking=-; providers=wandb; source-model=GLM-5.2%20%5BW%26B%5D) | 70.22 |
| 9 | Gemini 3.5 Flash (WolfBench Run: agent=hermes; version=v2026.5.16; timeout=3600s; thinking=medium; providers=google; source-model=Gemini%203.5%20Flash) | 69.29 |
| 10 | GPT-5.4 (WolfBench Run: agent=hermes; version=v2026.3.30; timeout=3600s; thinking=medium; providers=openai; source-model=GPT-5.4) | 67.42 |
Interactive version: theaggregate.ai/benchmark?slug=wolfbench-hermes · How It Works · Data refreshed daily, snapshot 2026-10-09.