TIDE (Evolving Documents, Closed-Book): leaderboard
TIDE (Sobhani et al., 2026) tests version resolution over temporally evolving documents: 3,050 expert-verified questions about 644 Bangladesh customs instruments issued from 1969 to 2025, whose correct answer depends on the version in force at the queried date. This board supplies no document, so answers rest on parametric knowledge; the score is strict accuracy in percent macro-averaged over eight task types, free-form answers judged by a three-model council with a hard date gate; higher is better.
Metric: Macro-averaged strict accuracy (%; mean over the eight task types; TIDE, 3,050 expert-verified questions over 644 Bangladesh customs instruments (1969-2025); free-form answers are judged by a council of Claude Opus 4.6, Gemini 3.5 Flash and GPT-5.1 by majority vote with a hard date gate, deterministic tasks by exact match; strict accuracy without the partial-rubric credit; thinking enabled, temperature 0; closed book: no document is supplied). Source: arxiv.org. Saturation forecast: Around 2029. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (Thinking) | 35.39 | #176 (GPT-5.2) |
| 2 | Gemini 3.5 Flash | 34.15 | #101 |
| 3 | Kimi K2.5 (Thinking) | 33.28 | #256 (Kimi K2.5) |
| 4 | Gemma 4 31B (Reasoning) | 32.04 | #394 (Gemma 4 31B) |
| 5 | GPT-OSS-120B | 30.76 | #700 |
| 6 | Qwen 3.6 35B A3B | 26.39 | #462 |
| 7 | Claude Sonnet 4.5 (Thinking) | 25.91 | #250 (Claude Sonnet 4.5) |
| 8 | GLM-4.6 | 25.54 | #486 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=tide-evolving-documents-closed-book · How It Works · Data refreshed daily, snapshot 2026-11-13.