List your company

Reasoning on Overall Combined Benchmarks

88.7AccuracyVanilla
19.22837.26455.373.336Apr 8, 2026Apr 14, 2026Apr 21, 2026Apr 27, 2026May 4, 2026May 10, 2026May 17, 2026

Evaluation results

MethodLinks
VanillaBackbone=Qwen3-14B2026.0488.75,062—
DTSRBackbone=Qwen3-14B2026.0488.73,246—
TALEBackbone=Qwen3-14B2026.0487.74,872—
PUMAModel=Qwen3-30B-A3B-Thinking2026.0582.5—28.2
Full CoTModel=Qwen3-30B-A3B-Thinking2026.0581.7—0
DEERModel=Qwen3-30B-A3B-Thinking2026.0581.3—22.4
DynasorModel=Qwen3-30B-A3B-Thinking2026.0579.7—3
CoDBackbone=Qwen3-14B2026.0471.12,077—
No-ThinkModel=Qwen3-30B-A3B-Thinking2026.0568.8—45.3
PUMAModel=Llama-3.1-Nemotron-Nano-8B2026.0564.7—20.1
Full CoTModel=Llama-3.1-Nemotron-Nano-8B2026.0564.3—0
DEERModel=Llama-3.1-Nemotron-Nano-8B2026.0561.2—-30.7
PUMAModel=DeepSeek-R1-Distill-Qwen-7B2026.0560.2—35.6
Plan&BudgetModel=Qwen3-30B-A3B-Thinking2026.0559.6—53.9
DynasorModel=Llama-3.1-Nemotron-Nano-8B2026.0558.3—-4.7
Full CoTModel=DeepSeek-R1-Distill-Qwen-7B2026.0558—0
DEERModel=DeepSeek-R1-Distill-Qwen-7B2026.0557.9—21.5
CCoTModel=Qwen3-30B-A3B-Thinking2026.0554.3—55.5
Plan&BudgetModel=Llama-3.1-Nemotron-Nano-8B2026.0552.1—18.6
CoDModel=Llama-3.1-Nemotron-Nano-8B2026.0551.7—31.7
CCoTModel=DeepSeek-R1-Distill-Qwen-7B2026.0550.4—58.6
CoDModel=DeepSeek-R1-Distill-Qwen-7B2026.0549.4—56.6
CCoTModel=Llama-3.1-Nemotron-Nano-8B2026.0548.8—34.1
Plan&BudgetModel=DeepSeek-R1-Distill-Qwen-7B2026.0547.8—47.9
DynasorModel=DeepSeek-R1-Distill-Qwen-7B2026.0545.9—36.6
CoDModel=Qwen3-30B-A3B-Thinking2026.0545.3—40.4
No-ThinkModel=DeepSeek-R1-Distill-Qwen-7B2026.0538.7—79.9
Ans. Conv.Model=DeepSeek-R1-Distill-Qwen-7B2026.0535.1—81.9
No-ThinkModel=Llama-3.1-Nemotron-Nano-8B2026.0532.8—-80.5
Ans. Conv.Model=Qwen3-30B-A3B-Thinking2026.0523.2—90.9
Ans. Conv.Model=Llama-3.1-Nemotron-Nano-8B2026.0521.9—83.7