Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Real-data Math

Benchmarks

Task NameDataset NameSOTA ResultTrend
MathReal-data Math Mistral outputs (3% verified data)
Original Accuracy50.46
6
MathReal-data Math Qwen outputs (3% verified data)
Original Accuracy55.1
6
MathReal-data Math Gemini outputs (3% verified data)
Original Accuracy66.35
6
MathReal-data Math GPT-5.4-mini (3% verified data)
Original Accuracy62.35
6
MathReal-data Math GPT-5.4 outputs (3% verified data)
Original Accuracy67.2
6
Showing 5 of 5 rows