Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Deep search on Average webw., hle, gaia
Loading...
9.87
Accuracy
Qwen3-8B + TEPOdense
6.126
7.098
8.07
9.042
Feb 2, 2026
Accuracy
Tools Usage
Updated 4d ago
Evaluation Results
Method
Method
Links
Accuracy
Tools Usage
Qwen3-8B + TEPOdense
Training Method=TEPOdense
2026.02
9.87
1.84
Qwen3-8B + AEPO
Training Method=AEPO
2026.02
9.69
3.21
Qwen3-8B + TEPOsparse
Training Method=TEPOsp...
2026.02
9.66
1.23
Qwen3-8B + ARPO
Training Method=ARPO
2026.02
9.48
4.63
Qwen3-8B + GRPO
Training Method=GRPO
2026.02
9.43
1.77
Qwen3-8B + SFT
Training Method=SFT
2026.02
9.06
5.65
Qwen3-8B
Training Method=Base
2026.02
6.27
1.14
Feedback
Search any
task
Search any
task