List your company

Planning on PlanBench

99.1AccuracyDeepSeek-R1 Distill
26.71645.50864.383.092May 1, 2026May 7, 2026May 13, 2026May 19, 2026May 25, 2026May 31, 2026Jun 6, 2026

Evaluation results

MethodLinks
DeepSeek-R1 DistillSize=70B2026.0699.1
TeamTR (Heterogeneous)Size=1.7B+8B+14B2026.0544.5
TeamTR (Homogeneous)Size=3×8B2026.0544.1
Qwen3-Base (DICE-PC)Size=3×4B2026.0642.7
Qwen3-235B-A22BSize=235B2026.0642.5
OpenAI o3Size=–2026.0642.1
Qwen3-Base (DICE-FT)Size=3×4B2026.0641.9
o4-miniSize=–2026.0541.2
OpenAI o3-miniSize=–2026.0641.2
Qwen3 (thinking)Size=32B2026.0540.3
Qwen3 (thinking)Size=32B2026.0640.3
Qwen3-Inst (Few-shot CoT)Size=4B2026.0640.1
Qwen3-Inst (SC@64)Size=14B2026.0639.7
Qwen3-A3BSize=30B2026.0539.1
DeepSeek-R1 DistillSize=70B2026.0539.1
Debate + JudgeSize=3×8B2026.0539.1
Qwen3-Base Debate+Judge (FT)Size=3×8B2026.0639.1
Qwen3-A3BSize=30B2026.0639.1
Role-playSize=3×8B2026.0538.7
Qwen3-Base Role-play (FT)Size=3×8B2026.0638.7
Qwen3-Base Debate+Judge (FT)Size=3×4B2026.0638.4
DebateSize=3×8B2026.0538.3
Qwen3-Base Debate (FT)Size=3×8B2026.0638.3
Qwen3-Inst Debate+Judge (prompt)Size=3×4B2026.0638.1
QwQSize=32B2026.0537.9
Qwen3-Inst (ReAct)Size=4B2026.0637.9
Qwen3-Base Role-play (FT)Size=3×4B2026.0637.9
QwQSize=32B2026.0637.9
Qwen3-Inst Role-play (prompt)Size=3×4B2026.0637.8
Qwen3-Base Debate (FT)Size=3×4B2026.0637.5
Qwen3-Inst Debate (prompt)Size=3×4B2026.0637.2
Llama 3.1-Base (DICE-PC)Size=3×8B2026.0637.1
Qwen3-Inst (ToT)Size=4B2026.0637
Qwen3-Inst (SC@64)Size=4B2026.0636.3
Qwen3-Base-DAPOSize=14B2026.0635.6
Qwen3-Base-GRPOSize=14B2026.0634.8
GPT-4o-miniSize=–2026.0534.7
GPT-4o-miniSize=–2026.0634.7
Qwen3-BaseSize=32B2026.0634.3
Qwen2.5-BaseSize=72B2026.0633.7
DAPOSize=8B2026.0533.5
Qwen3-Base-DAPOSize=8B2026.0633.5
GRPOSize=8B2026.0533.1
Qwen3-Base-GRPOSize=8B2026.0633.1
Qwen3-Base-PPOSize=8B2026.0632.6
Qwen3-BaseSize=14B2026.0629.7
Qwen3-Inst (Zero-shot CoT)Size=4B2026.0629.5