ResearchBenchmarksScientific Reasoning on FrontierSciFollow55.7Pass@1CoNL6.319.12531.9544.775Jan 29, 2026Evaluation ResultsMethodMethodLinksPass@1CoNLBackbone=Qwen3-8B, Tra...Backbone=Qwen3-8B, Training Required=true2026.0155.7CoNLBackbone=Qwen3-4B-Inst...Backbone=Qwen3-4B-Instruct, Training Required=true2026.0127.5CoNLBackbone=Llama-3.1-8B,...Backbone=Llama-3.1-8B, Training Required=true2026.0110.2CoNLBackbone=Llama-3.2-3B,...Backbone=Llama-3.2-3B, Training Required=true2026.018.2