FRED-SemBench - LLM Only 22-Question Iris-Supported Slice: leaderboard
FRED-SemBench tests whether an unaided language model can interpret economic data correctly, including values, units and when clarification is needed. Deterministic strict value/unit/clarification accuracy (%).
Metric: Deterministic strict value/unit/clarification accuracy (%). Source: aclanthology.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (medium) | 18.18 | #22 (GPT-5.6 Sol) |
Interactive version: theaggregate.ai/benchmark?slug=fred-sembench-llm-only-22-question-iris-supported-slice · How It Works · Data refreshed daily, snapshot 2026-10-29.