TS-Skill (MCQ) - SK1+SK2+SK3: leaderboard

TS-Skill questions (Han et al., UCLA and Carnegie Mellon University, 2026) about synthetic, timestamped time series that need all three skills at once: temporal scale selection, temporal localization and cross-interval integration. Multiple-choice protocol: 249 questions answered with a single option letter, unparseable replies scored 0; macro-F1 on a 0-1 scale, higher is better.

Metric: MCQ macro-F1 (0-1; four-option multiple choice, letter-only answers, unparseable answers score 0, 249 questions). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.

Top models

#ModelScoreOverall rank
1GPT-4o0.77#741
2Claude Sonnet 4.50.77#244
3Qwen 3 32B (Thinking)0.71#960 (Qwen 3 32B)
4Qwen 2.5 VL 7B0.59#1796
5Qwen 2.5 14B Instruct0.56#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-mcq-sk1-plussk2-plussk3 · How It Works · Data refreshed daily, snapshot 2026-11-12.