Reasoning over conflicting evidence on SEAL-0
45.9AccuracyROMA
Evaluation Results
| Method | Links | |
|---|---|---|
| ROMASource type=Open-source, Search=true, backbone=GLM-4.62026.02 | 45.9 | |
| Kimi-ResearcherSource type=Open-source, Search=true2026.02 | 36 | |
| Perplexity Deep ResearchSource type=Closed-source, Search=true2026.02 | 31.5 | |
| Grok-4Source type=Closed-source, Search=true2026.02 | 20.7 | |
| Gemini 2.5 ProSource type=Closed-source, Search=true2026.02 | 19.8 | |
| o3-proSource type=Closed-source, Search=true2026.02 | 18.9 | |
| o3Source type=Closed-source, Search=true2026.02 | 15.3 | |
| GLM-4.6Source type=Open-source, Search=true2026.02 | 14.5 | |
| Gemini 2.5 FlashSource type=Closed-source, Search=true2026.02 | 13.5 | |
| Perplexity Sonar Reasoning ProSource type=Closed-source, Search=true2026.02 | 13.5 | |
| Open Deep SearchSource type=Open-source, Search=true, backbone=DeepSeek-R12026.02 | 9.9 | |
| Grok-3Source type=Closed-source, Search=true2026.02 | 5.4 | |
| Qwen3-235B-A22BSource type=Open-source, Search=true2026.02 | 5.4 | |
| DeepSeek-R1Source type=Open-source, Search=true2026.02 | 4.5 |