ThermEval - Temperature Estimation (Coordinates): leaderboard
Metric: Mean absolute error (degrees Celsius, 0 or more) on Task 6, the temperature at given pixel coordinates of a ThermEval-D image with axes and a colorbar (2,400 questions); zero-shot with fixed one-word or one-number prompts, a single forward pass; Gemini 2.5 models parse free-form answers into the label or number only when the output deviates from the format; lower is better. Source: arxiv.org. Saturation forecast: Around April 2027. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 1.94 | #77 |
| 2 | Gemini 3 Flash | 1.96 | #93 |
| 3 | InternVL3-14B | 2.6 | #494 |
| 4 | InternVL3-38B | 2.97 | #395 |
| 5 | Llama 3.2 11B Instruct | 3 | #1112 |
| 6 | Qwen 2.5 VL 7B Instruct | 3.21 | #643 |
| 7 | Qwen 2 VL 7B Instruct | 3.28 | #816 |
| 8 | Gemini 2.5 Pro | 3.47 | #145 |
| 9 | Qwen 2.5 VL 32B Instruct | 3.57 | #443 |
| 10 | Gemini 2.5 Flash | 4.34 | #237 |
| 11 | MiniCPM-V-2.6 | 9.16 | #825 |
| 12 | InternVL3-8B | 31.92 | #606 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=thermeval-temperature-estimation-coordinates · How It Works · Data refreshed daily, snapshot 2026-10-11.