JuICE: leaderboard
As large language models (LLMs) are increasingly deployed to users around the world, they are integrated into everyday tasks across diverse cultural contexts, from drafting personal communications to brainstorming creative ideas. These tasks are inherently cultural: they require contextual appropriateness, symbolic resonance, and tacit cultural expectations that native speakers draw on instinctive
Metric: F1 Score (self-reported). Source: benchmarklist.com. Saturation forecast: Around July 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 56.67 | #58 |
| 2 | Gemini 3 Flash (Preview) | 52.14 | #82 |
| 3 | GPT-5.5 | 49.67 | #31 |
| 4 | Claude Haiku 4.5 | 40.38 | #280 |
| 5 | Claude Opus 4.7 | 36.93 | #49 |
| 6 | Gemma 4 31B (IT) | 36.23 | #176 |
| 7 | GPT-OSS-120B | 34.7 | #341 |
| 8 | Qwen 3 30B A3B 2507 Instruct | 33.74 | #488 |
| 9 | GPT-5.4 Mini | 20.82 | #212 |
| 10 | Llama 4 Scout Instruct | 17.32 | #539 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=juice · How It Works · Data refreshed daily, snapshot 2026-10-20.