List your company

Scientific Reasoning on GPQA-D (Acc, Tok, CR)

69.5AccuracyVanilla
36.11644.78353.4562.117Apr 19, 2026

Evaluation results

MethodLinks
VanillaBackbone=Gemini2.5-Flash2026.0469.54,565100
DynasorBackbone=Gemini2.5-Flash2026.0469.54,09189.6
TRACEBackbone=Gemini2.5-Flash2026.0469.43,26271.4
DEERBackbone=Gemini2.5-Flash2026.0469.13,39774.4
TALEBackbone=Gemini2.5-Flash2026.0468.23,59778.8
NoThinkBackbone=Gemini2.5-Flash2026.0467.34,36595.6
VanillaBackbone=Qwen3-8B2026.0459.19,398100
TRACEBackbone=Qwen3-8B2026.0456.65,62959.9
VanillaBackbone=Qwen3-4B2026.0455.18,611100
TRACEBackbone=Qwen3-4B2026.04555,71367.3
DynasorBackbone=Qwen3-8B2026.0454.65,73261
DEERBackbone=Qwen3-4B2026.0454.57,04281.8
DEERBackbone=Qwen3-8B2026.0453.27,27977.5
DynasorBackbone=Qwen3-4B2026.0452.14,81455.9
TRACEBackbone=R1-Distilled-Llama-8B2026.0451.54,64475.1
NoThinkBackbone=Qwen3-8B2026.04511,53816.4
VanillaBackbone=R1-Distilled-Llama-8B2026.04516,184100
NoThinkBackbone=R1-Distilled-Llama-8B2026.0450.55,84994.6
TALEBackbone=Qwen3-8B2026.04504,24545.2
NoThinkBackbone=Qwen3-4B2026.04491,65819.3
TALEBackbone=Qwen3-4B2026.0447.95,80767.4
DynasorBackbone=R1-Distilled-Llama-8B2026.0443.64,01965
DEERBackbone=R1-Distilled-Llama-8B2026.04434,73176.5
TALEBackbone=R1-Distilled-Llama-8B2026.0437.44,53173.3