ABRA: leaderboard
Existing medical-agent benchmarks deliver imaging as pre-selected samples, never as an environment the agent must navigate. We introduce ABRA, a radiology-agent benchmark in which the agent operates an OHIF viewer and an Orthanc DICOM server through twenty-one function-calling tools that span slice navigation, windowing, series selection, pixel-coordinate annotation, and structured reporting. ABRA
Metric: Overall Avg (imported). Source: benchmarklist.com. Saturation forecast: Around August 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 70 | #90 |
| 2 | GPT-5.4 | 70 | #79 |
| 3 | Qwen 3.5 397B A17B | 70 | #146 |
| 4 | GPT-5.4 Nano | 69 | #328 |
| 5 | Gemma 4 31B (IT) | 68 | #176 |
| 6 | Mistral Large 3 | 67 | #401 |
| 7 | Gemini 3 Flash | 66 | #99 |
| 8 | Ministral 3 14B | 66 | #650 |
| 9 | Kimi K2.5 | 64 | #144 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=abra · How It Works · Data refreshed daily, snapshot 2026-10-20.