Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Downstream task performance on SkillsBench
Loading...
51.1
Pass Rate
Gold Skills
21.044
28.847
36.65
44.453
Jun 30, 2026
Pass Rate
Token Usage (M)
Updated 24d ago
Evaluation Results
Method
Method
Links
Pass Rate
Token Usage (M)
Gold Skills
Backbone Model=GPT-5.2...
2026.06
51.1
1.12
Gold Skills
Backbone Model=Gemini-...
2026.06
48.4
1.18
SkillComposer
Backbone Model=GPT-5.2...
2026.06
45.3
1.03
Retrieval (oracle)
Backbone Model=GPT-5.2...
2026.06
44
1.13
Retrieval (top-3)
Backbone Model=GPT-5.2...
2026.06
44
1.09
SkillComposer
Backbone Model=Gemini-...
2026.06
44
1.08
Retrieval (oracle)
Backbone Model=Gemini-...
2026.06
42.2
1.19
Retrieval (top-3)
Backbone Model=Gemini-...
2026.06
41.8
1.14
All Skills
Backbone Model=Gemini-...
2026.06
38.7
1.33
All Skills
Backbone Model=GPT-5.2...
2026.06
29.3
1.27
No Skills
Backbone Model=Gemini-...
2026.06
25.8
0.99
No Skills
Backbone Model=GPT-5.2...
2026.06
22.2
0.94
Feedback
Search any
task
Search any
task