Horizon - Medium — leaderboard
Metric: Task Pass Rate (%). Source: www.orinlabs.ai. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | OpenClaw (LCM) - claude-opus-4.8 | 70.77 |
| 2 | OpenClaw (LCM) - gpt-5.5 | 61.54 |
| 3 | OpenClaw (LCM) - gemini-3.5-flash | 60 |
| 4 | RLM - claude-opus-4.8 | 58.46 |
| 5 | OpenClaw (LCM) - gemini-3.1-pro-preview | 55.38 |
| 6 | RLM - gpt-5 | 53.85 |
| 7 | RLM - claude-sonnet-4.6 | 53.85 |
| 8 | Codex - gpt-5-codex | 46.15 |
| 9 | OpenClaw (LCM) - deepseek-v4-pro | 43.08 |
| 10 | Claude Code - claude-opus-4.8 | 41.54 |
Interactive version: theaggregate.ai/benchmark?slug=horizon-medium · How the rankings work · Data refreshed daily, snapshot 2026-07-22.