CHI-Bench — leaderboard
Healthcare workflow agent benchmark across prior authorization, utilization management, and care management, with agents operating simulated clinical apps through MCP-style tools.
Metric: Overall Pass@1 (%). Source: actava.ai. Status: saturation imminent. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.8 | 37.3 |
| 2 | Claude Opus 4.6 | 28 |
| 3 | Claude Sonnet 4.6 | 26.2 |
| 4 | Claude Opus 4.7 | 24.4 |
| 5 | GPT-5.5 | 20.9 |
| 6 | Claude Sonnet 5 | 20 |
| 7 | GLM-5.1 | 18.7 |
| 8 | GLM-5.2 | 18.7 |
| 9 | GPT-5.4 | 16 |
| 10 | Kimi K2.6 | 15.6 |
| 11 | DeepSeek V4 Pro | 14.2 |
| 12 | Gemini 3 Flash | 12.5 |
| 13 | GPT-5.4 Mini | 8.4 |
| 14 | Gemini 3.1 Pro (Preview) | 7.1 |
| 15 | Claude Haiku 4.5 | 6.2 |
Interactive version: theaggregate.ai/benchmark?slug=chi-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.