LaViSA - Ellipsis: leaderboard
LaViSA (2026) shows a vision-language model a structurally ambiguous sentence and a generated image, and asks which of two or three disambiguated readings the image depicts. This board is per-trial accuracy in percent on the ellipsis category (chance 50); higher is better.
Metric: Per-trial accuracy (%) on the ellipsis category; task: pick which disambiguated reading of a structurally ambiguous sentence a generated image depicts (2 or 3 readings per sentence, 100 sentences per category); chance 50; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 83.2 | #94 |
| 2 | GPT-5.2 | 71.1 | #176 |
| 3 | Gemini 3.1 Flash Lite | 70.8 | #364 |
| 4 | Qwen 3 VL 8B Instruct | 66.8 | #981 |
| 5 | Qwen 3 VL 32B (Thinking) | 65.8 | #559 |
| 6 | Qwen 3 VL 32B Instruct | 63.9 | #585 |
| 7 | Gemma 3 12B (IT) | 61.9 | #1837 |
| 8 | Gemma 3 27B (IT) | 61.4 | #1289 |
| 9 | Qwen 3 VL 4B Instruct | 61.4 | #1279 |
| 10 | Qwen 3 VL 8B (Thinking) | 59.4 | #887 |
| 11 | Gemma 3 4B (IT) | 54.5 | #3831 |
| 12 | Qwen 3 VL 4B (Thinking) | 51 | #1296 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=lavisa-ellipsis · How It Works · Data refreshed daily, snapshot 2026-11-13.