aiXamine - OOD Robustness: leaderboard
aiXamine OOD Robustness service score: stable task performance under word-level and style-shifted (Bible, romantic, Shakespeare, tweet) inputs (aiXamine, arXiv 2608.20554v1, Appendix Table 7). Normalized 0-100 score aggregated from the service's tests; higher is better. 42 models evaluated under identical black-box conditions via OpenRouter or containerized vLLM.
Metric: Service score (0-100; normalized pass rate). Source: arxiv.org. Saturation forecast: Estimated already saturated. 42 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 91.6 | #90 |
| 2 | Grok 4 Fast | 90 | #266 |
| 3 | Grok 4.1 Fast | 89.8 | #227 |
| 4 | Grok 3 Mini | 89.6 | #289 |
| 5 | GPT-5 Mini | 89.5 | #193 |
| 6 | Gemma 4 26B A4B | 89.5 | #360 |
| 7 | Gemma 4 31B | 89.1 | #212 |
| 8 | Gemma 2 27B | 88.6 | #803 |
| 9 | Gemini 2.0 Flash (001) | 88.6 | #365 |
| 10 | Llama 3.3 70B | 88.6 | #578 |
| 11 | Llama 4 Scout | 88.4 | #648 |
| 12 | Gemini 2.5 Flash | 88.3 | #255 |
| 13 | Claude Haiku 4.5 | 88.1 | #287 |
| 14 | Phi-4 | 87.3 | #707 |
| 15 | Claude 3.5 Haiku | 87 | #571 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aixamine-ood-robustness · How It Works · Data refreshed daily, snapshot 2026-11-05.