← All incidents
✓ ResolvedConfidence: Low
Responses to fixed prompts no longer match this model’s own history (1344 recent vs 11206 baseline responses). Biggest shifts: output length on “Name a famous scientist. Reply with only the name.”; reasoning tokens on “Name a famous scientist. Reply with only the name.”; output length on “Pick a number between 1 and 1000. Reply with only the number.”
Oct 17change began ≈ 1:07 AM UTC
Oct 18incident opened 7:30 AM UTC
Oct 18resolved 10:51 AM UTC
7.2e-3adjusted p
What this means
Raven Alpha compares how the model answers a fixed set of open-ended prompts with its own history. A persistent, significant shift means the model, its configuration or its serving path changed, even when every scored test still passes.
Opened Oct 18, 2026, 7:30 AM UTC UTC. Resolved Oct 18, 2026, 10:51 AM UTC UTC: the series returned to its baseline. Methodology.