TS-Skill (Free-Form) - SK2+SK3: leaderboard
TS-Skill questions (Han et al., UCLA and Carnegie Mellon University, 2026) about synthetic, timestamped time series that combine temporal localization with cross-interval integration, such as aggregating values over intervals anchored to events or timestamps. Free-form protocol: 342 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 342 questions). Source: arxiv.org. Saturation forecast: Around March 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 0.58 | #741 |
| 2 | O3 Mini | 0.52 | #536 |
| 3 | Claude Sonnet 4.5 | 0.5 | #244 |
| 4 | Qwen 3 32B (Thinking) | 0.45 | #960 (Qwen 3 32B) |
| 5 | Qwen 2.5 14B Instruct | 0.42 | #1745 |
| 6 | Qwen 2.5 VL 7B | 0.39 | #1796 |
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-sk2-plussk3 · How It Works · Data refreshed daily, snapshot 2026-11-12.