MTBBench-Longitudinal - Outcome: leaderboard
Outcome-prediction questions of the MTBBench longitudinal track (Vasilev et al., 2025): from a cancer patient's evolving clinical timeline and genomic data the agent predicts survival over a stated horizon. Accuracy in percent, no tools; higher is better.
Metric: Accuracy (%; outcome-prediction questions over the evolving clinical timeline; multi-turn file access, no tools). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct [4bit] | 73.2 | #1221 |
| 2 | GPT-4o (2024-08-06) | 72.9 | #731 |
| 3 | O4 Mini (2025-04-16) | 66 | #362 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mtbbench-longitudinal-outcome · How It Works · Data refreshed daily, snapshot 2026-11-12.