BehaviorBench (Behavioral Traces) - Trade Amount Error (RetrievalGen): leaderboard
Metric: Median absolute error of the predicted trade amount (paper units, unbounded); history interface RetrievalGen (behaviorally similar examples retrieved from a disjoint support pool of wallets); 2,000 pseudonymous prediction-market wallets reconstructed from public Polymarket metadata and Polygon on-chain logs, about 59,560 chronologically held-out test transactions; unparsable outputs count as failures; context-only baseline 31.79; lower is better. Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 3 12B (IT) | 8.29 |
| 2 | Claude 3.7 Sonnet | 9.76 |
| 3 | GPT-5.4 | 10 |
| 4 | GPT-5.4 Nano | 10.22 |
| 5 | Llama 4 Scout Instruct | 10.81 |
| 6 | GPT-5.4 Mini | 11.13 |
| 7 | NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 | 11.53 |
| 8 | GPT-OSS-20B | 14.5 |
| 9 | Gemma 3 27B (IT) | 14.65 |
| 10 | Ministral-3-14B-Instruct-2512 | 21.8 |
| 11 | Claude Opus 4 | 22.67 |
Interactive version: theaggregate.ai/benchmark?slug=behaviorbench-behavioral-traces-trade-amount-error-retrievalgen · How It Works · Data refreshed daily, snapshot 2026-09-29.