SuperCLUE-Ondevice - Overall: leaderboard
SuperCLUE is the Chinese-language evaluation programme run by CLUE; its specialised boards each cover one professional domain and are graded by the organisers rather than self-reported. This board covers on-device models, overall — the board exists because the phone-sized tier is not comparable to the frontier one.
Metric: Score. Source: www.superclueai.com. Saturation forecast: Around September 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.6 Flash | 93.64 | #51 |
| 2 | Seed 2.1 Pro | 92.99 | #67 |
| 3 | Qwen 3.8 Max | 92.25 | #35 |
| 4 | Qwen 3.5 9B | 87.82 | #379 |
| 5 | GLM-5V Turbo | 87.75 | #166 |
| 6 | Qwen 3.5 4B | 85.16 | #475 |
| 7 | gemma-4-E4B-it | 72.96 | #566 |
| 8 | Gemma 4 E2B (IT) | 66.45 | #781 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=superclue-ondevice-overall · How It Works · Data refreshed daily, snapshot 2026-10-20.