ThermEval - Temperature Estimation at 10 ft: leaderboard

Metric: Mean absolute error (degrees Celsius, 0 or more) on Task 7, body-region temperature from ThermEval-D images taken at 10 ft (333 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: Not forecast. 23 models tracked.

Top models

#ModelScoreOverall rank
1InternVL3-8B0.76#606
2Qwen 2.5 VL 7B Instruct0.87#643
3InternVL3-14B0.89#494
4Qwen 2.5 VL 32B Instruct0.89#443
5Qwen 2 VL 7B Instruct0.89#816
6Gemini 3 Pro0.9#77
7InternVL3-38B0.98#395
8Gemini 3 Flash1.21#93
9Llama 3.2 11B Instruct1.48#1112
10Gemini 2.5 Pro2.66#145
11Gemini 2.5 Flash2.83#237
12MiniCPM-V-2.68.56#825

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-at-10-ft · How It Works · Data refreshed daily, snapshot 2026-10-11.