List your company

Q&A Reasoning on GPQA (Strict Accuracy, Token Metrics)

45.46Strict Match AccuracyCactus 0.75
38.627240.401142.17543.9489Apr 5, 2026

Evaluation results

MethodLinks
Cactus 0.75m (draft length)=20, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0445.466.46-46
Cactus 1.0m (draft length)=20, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0445.466.74-50
Cactus 0.75m (draft length)=10, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0444.955.14-50
Cactus 1.0m (draft length)=10, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0443.435.16-51
SpSm (draft length)=20, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0442.934.23—
Cactus 0.75m (draft length)=10, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0441.425.33-47
TASm (draft length)=10, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0441.415.17-42
VerifierVerifier model=Qwen 3 8B2026.0441.07——
SpSm (draft length)=10, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0440.913.7—
SpSm (draft length)=20, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0440.913.84—
TASm (draft length)=20, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0440.46.41-46
VerifierVerifier model=Qwen 3 14B2026.0440.07——
Cactus 0.75m (draft length)=20, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0440.016.89-47
Cactus 1.0m (draft length)=20, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0439.97.05-49
Cactus 1.0m (draft length)=10, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0439.395.44-48
SpSm (draft length)=10, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0439.393.37—
TASm (draft length)=20, Verifier model=Qwen 3 8B, Drafter model=Qwen 3 0.6B2026.0438.896.68-46
TASm (draft length)=10, Verifier model=Qwen 3 14B, Drafter model=Qwen 3 0.6B2026.0438.894.99-46