WearableQA - Overall: leaderboard
WearableQA (arXiv 2609.05405v1): 4,084 ten-option multiple-choice questions over wearable time series, blood biomarkers and demographics of 200 real users with up to 500 days of data. Overall accuracy with chain-of-thought prompting at temperature 0, chance 10%, unparseable answers counted wrong (Table 1). Higher is better.
Metric: Accuracy (%; 10-option multiple choice, chain-of-thought). Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 72.9 | #64 |
| 2 | Claude Opus 4.6 | 60.2 | #67 |
| 3 | Claude Sonnet 4 | 53.2 | #207 |
| 4 | GPT-5.4 | 51.2 | #83 |
| 5 | Gemini 2.5 Pro | 50.6 | #159 |
| 6 | Gemma 4 26B A4B | 42.5 | #360 |
| 7 | GPT-4o | 34.7 | #357 |
| 8 | Mistral Small 3.1 | 29.3 | #623 |
| 9 | Llama 3.3 70B | 29.2 | #578 |
| 10 | Gemma 3 27B | 26.2 | #606 |
| 11 | Gemma 3 12B | 24.6 | #674 |
| 12 | Llama 3.1 8B | 23.8 | #1130 |
| 13 | Gemma 3 4B | 19.8 | #1088 |
| 14 | Llama 3.2 3B | 19.6 | #1376 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=wearableqa-overall · How It Works · Data refreshed daily, snapshot 2026-11-05.