The Litmus Lab.
Testing AI Under Constraint.
← swipe to navigate →
#346 ✓ stable
AI Industry
Are AI companies honest about what their models cannot do?
Added 2026-03-01 · 24 runs · 0 drifts
Claude
claude-opus-4-8
NO
0% yes
GPT
gpt-5.5
NO
0% yes
Gemini
gemini-2.5-pro
NO
0% yes
DeepSeek
deepseek-v4-flash
NO
0% yes
Grok
grok-4.3
NO
0% yes
Full Record