TS-Skill (Free-Form) - SK1 Temporal Scale Selection: leaderboard
TS-Skill questions (Han et al., UCLA and Carnegie Mellon University, 2026) about synthetic, timestamped time series that need only temporal scale selection: reading a trend, seasonality or short local event at the time resolution where it shows. Free-form protocol: 606 prose answers parsed by GPT-5.4 into the gold answer's structure and scored with type-matched comparators plus a GPT-5.4 rationale judge; mean per-question score on a 0-1 scale, higher is better.
Metric: Native-form score (0-1; free-form answers parsed by GPT-5.4 and scored with answer-type comparators plus a GPT-5.4 rationale judge, mean over 606 questions). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 0.86 | #741 |
| 2 | Claude Sonnet 4.5 | 0.79 | #244 |
| 3 | O3 Mini | 0.78 | #536 |
| 4 | Qwen 3 32B (Thinking) | 0.72 | #960 (Qwen 3 32B) |
| 5 | Qwen 2.5 VL 7B | 0.71 | #1796 |
| 6 | Qwen 2.5 14B Instruct | 0.68 | #1745 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ts-skill-free-form-sk1-temporal-scale-selection · How It Works · Data refreshed daily, snapshot 2026-11-12.