TFRBench - Hierarchical Sales - Logic-to-Number Consistency: leaderboard
Metric: Mean LLM-judge score (1-5) for logic-to-number consistency of the model's chain-of-thought forecasting reasoning on the TFRBench daily hierarchical sales (GIFT-Eval, sales) windows, judged by Gemini 3 Pro against the benchmark's verified reference reasoning (w/ Reasoning setting: the model reasons step by step from the history window, then forecasts; no tools; mean of three runs); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 4.75 |
| 2 | Gemini 2.5 Pro | 4.73 |
| 3 | Gemini 3 Pro | 4.34 |
| 4 | Claude Sonnet 4.5 | 3.65 |
| 5 | Claude Sonnet 4 | 3.33 |
Interactive version: theaggregate.ai/benchmark?slug=tfrbench-hierarchical-sales-logic-to-number-consistency · How It Works · Data refreshed daily, snapshot 2026-10-07.