Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multiple-choice Question Answering on GPQA (accuracy)
Loading...
30.35
Accuracy (%)
DS2-INSTRUCT
7.262
13.256
19.25
25.244
Mar 13, 2026
Accuracy (%)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy (%)
DS2-INSTRUCT
Model Family=Llama3
2026.03
30.35
DS2-INSTRUCT
Model Family=Mistral
2026.03
30.16
Self-Instruct
Model Family=Qwen2.5
2026.03
27.18
DS2-INSTRUCT
Model Family=Qwen2.5
2026.03
26.35
Zero-Shot
Model Family=Qwen2.5
2026.03
26.32
InstructMix
Model Family=Qwen2.5
2026.03
26.09
ExploreInstruct
Model Family=Qwen2.5
2026.03
25.93
InstructMix
Model Family=Mistral
2026.03
23.91
Self-Instruct
Model Family=Llama3
2026.03
19.87
ExploreInstruct
Model Family=Mistral
2026.03
19.47
ExploreInstruct
Model Family=Llama3
2026.03
18.24
InstructMix
Model Family=Llama3
2026.03
17.52
Zero-Shot
Model Family=Llama3
2026.03
12.51
Zero-Shot
Model Family=Mistral
2026.03
8.64
Self-Instruct
Model Family=Mistral
2026.03
8.15
Feedback
Search any
task
Search any
task