OmniEarth - Long-Term Trend Reasoning: leaderboard
Metric: CIDEr against reference descriptions, as the paper prints it (scaled by 1000), Long-Term Trend Reasoning (LTR) task of OmniEarth, zero-shot with each model's default inference settings; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-8B | 109.2 | #606 |
| 2 | Qwen 2.5 VL 72B Instruct | 100.7 | #364 |
| 3 | Gemini 2.0 Flash | 87.9 | #331 |
| 4 | GLM-4.6V | 82.7 | #309 |
| 5 | GPT-4o | 68.9 | #333 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=omniearth-long-term-trend-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-11.