List your company

Scientific Reasoning on GPQA (pass@1)

70.2pass@1EvoEnv
10.19225.77141.3556.929Apr 9, 2026Apr 14, 2026Apr 20, 2026Apr 26, 2026May 2, 2026May 8, 2026May 14, 2026

Evaluation results

MethodLinks
EvoEnvBase Model=Qwen3-4B-Thinking-25072026.0570.2
RLVEBase Model=Qwen3-4B-Thinking-25072026.0567.7
R-ZeroBase Model=Qwen3-4B-Thinking-25072026.0567.2
RLVEBase Model=Nemotron-Cascade-8B2026.0566.2
EvoEnvBase Model=Nemotron-Cascade-8B2026.0565.7
UntrainedBase Model=Qwen3-4B-Thinking-25072026.0565.2
UntrainedBase Model=Nemotron-Cascade-8B2026.0565.2
DAPOBase Model=Qwen3-4B-Thinking-25072026.0564.6
DAPOBase Model=Nemotron-Cascade-8B2026.0561.6
R-ZeroBase Model=Nemotron-Cascade-8B2026.0558.1
RLVEBase Model=Qwen3-4B-Instruct-25072026.0557.1
TTVSBackbone=Qwen3-8B2026.0456.1
DAPOBase Model=Qwen3-4B-Instruct-25072026.0556.1
EvoEnvBase Model=Qwen3-4B-Instruct-25072026.0555
R-ZeroBase Model=Qwen3-4B-Instruct-25072026.0554
TTRLBackbone=Qwen3-8B2026.0453
UntrainedBase Model=Qwen3-4B-Instruct-25072026.0552.5
TTVSBackbone=Qwen2.5-Instruct-7B2026.0451.5
TTVSBackbone=Qwen3-4B2026.0448.9
TTRLBackbone=Qwen2.5-Instruct-7B2026.0448.5
Qwen3-8BTest-time method=Base2026.0448
TTRLBackbone=Qwen3-4B2026.0443
TTVSBackbone=LLaMA3.1-8B2026.0437.3
TTVSBackbone=Qwen2.5-Instruct-3B2026.0436.4
Qwen2.5-Instruct-7BTest-time method=Base2026.0436.4
TTRLBackbone=LLaMA3.1-8B2026.0434.1
TTRLBackbone=Qwen2.5-Instruct-3B2026.0433.3
LLaMA3.1-8BTest-time method=Base2026.0430.8
Qwen2.5-Instruct-3BTest-time method=Base2026.0429.3
OpenReasoner-Zero-7BLabeled Data Size=129K2026.0428.4
TTVSBackbone=Qwen2.5-Instruct-1.5B2026.0428.3
SimpleRL-Zero-7BLabeled Data Size=8.9K2026.0427.6
TTVSBackbone=Qwen3-1.7B-Instruct2026.0426.3
Qwen3-4BTest-time method=Base2026.0426.3
DeepSeek-R1-Distill-7BLabeled Data Size=800K2026.0425.7
TTRLBackbone=Qwen2.5-Instruct-1.5B2026.0425.3
TTRLBackbone=Qwen3-1.7B-Instruct2026.0422.2
TTVSBackbone=LLaMA-3.2-3B-Instruct2026.0417.7
Qwen2.5-Instruct-1.5BTest-time method=Base2026.0416.7
TTRLBackbone=LLaMA-3.2-3B-Instruct2026.0416.1
DeepSeek-R1-Distill-1.5BLabeled Data Size=800K2026.0414.6
Qwen3-1.7B-InstructTest-time method=Base2026.0413.1
LLaMA-3.2-3B-InstructTest-time method=Base2026.0412.5