RSHR-Bench - Open Ended - MTEM@80: leaderboard
Primary RSHR-Bench remote-sensing high-resolution evaluation. Open Ended - MTEM@80. Multiple choice and open-ended judge-threshold protocols remain separate. Captioning uses the 1,000-image main-paper subset. Text-only baselines retain their input configuration.
Metric: Reported score (0–100). Source: arxiv.org. Saturation forecast: Around January 2027. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 72.8 | #94 |
| 2 | GPT-4o Mini | 72.5 | #588 |
| 3 | GPT-4o | 68.9 | #334 |
| 4 | Gemini 2.5 Pro | 20.6 | #150 |
| 5 | Qwen 2.5 VL 7B | 2.6 | #632 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=rshr-bench-open-ended-mtem-80 · How It Works · Data refreshed daily, snapshot 2026-10-16.