List your company

Mathematical Reasoning on MATH-500 (Accuracy, Token Count, Calibration Rate)

93.5AccuracyVanilla
68.85275.25181.6588.049Aug 13, 2025Sep 17, 2025Oct 22, 2025Nov 27, 2025Jan 1, 2026Feb 5, 2026Mar 13, 2026

Evaluation results

MethodLinks
VanillaBackbone=Ministral-3-8B-Reasoning-25122026.0393.56,212100
VanillaBackbone=Ministral-3-14B-Reasoning-25122026.03936,385100
OriginalBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0892.24,223—
VanillaBackbone=Qwen3-14B2026.03924,598100
LCPOBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0891.42,033—
DASTBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0891.23,563—
VanillaBackbone=Qwen3-8B2026.0391.15,037100
TERMINATORBackbone=Qwen3-8B2026.0390.72,42545.1
TERMINATORBackbone=Qwen3-14B2026.0390.72,26146.8
TERMINATORBackbone=Ministral-3-14B-Reasoning-25122026.0390.22,94643.9
TrEffBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0890.22,413—
Thought-CalibBackbone=Qwen3-8B2026.0390.14,37293.9
CoDBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0890.062,778—
Thought-CalibBackbone=Qwen3-14B2026.0389.83,77892
L1-ExactBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0889.83,555—
TERMINATORBackbone=Ministral-3-8B-Reasoning-25122026.0389.12,86347.8
L1-MaxBackbone=DeepSeek-R1-Distill-Qwen-7B2025.0888.82,016—
DynasorBackbone=Ministral-3-8B-Reasoning-25122026.0388.12,96756.8
Thought-CalibBackbone=Ministral-3-8B-Reasoning-25122026.0387.75,69587.8
Thought-CalibBackbone=Ministral-3-14B-Reasoning-25122026.0387.35,86095.9
DynasorBackbone=Ministral-3-14B-Reasoning-25122026.0386.33,24055.5
NoThinkingBackbone=Qwen3-14B2026.0384.178617.5
NoThinkingBackbone=Ministral-3-8B-Reasoning-25122026.0383.21,90828.1
LCPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0883.22,397—
OriginalBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.08835,665—
TrEffBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0882.82,813—
L1-MaxBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0882.23,374—
CoDBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0881.84,788—
L1-ExactBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.0881.83,335—
DEERBackbone=Qwen3-14B2026.0380.92,50156.2
NoThinkingBackbone=Qwen3-8B2026.0380.780916.1
DEERBackbone=Qwen3-8B2026.0379.92,60252
DynasorBackbone=Qwen3-14B2026.0379.61,70242.4
NoThinkingBackbone=Ministral-3-14B-Reasoning-25122026.0379.153511.7
DynasorBackbone=Qwen3-8B2026.0378.31,85041
DEERBackbone=Ministral-3-8B-Reasoning-25122026.03713,79160.3
DEERBackbone=Ministral-3-14B-Reasoning-25122026.0369.84,27974.6