Evidence · first-party tested

The model consistently missed the requested angry, guarded interrogation expression, instead producing a calm or neutral face in both portrait references.

✗ FailedLeonardo AI
The observation

The model consistently missed the requested angry, guarded interrogation expression, instead producing a calm or neutral face in both portrait references.

Criterion: Expression accuracy ·
Query this
get_evidence({
  tool: "leonardo-ai"
})
MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Real outputs, no retouching · every cell queryable via API & MCP · aidemos.com