MartingaleBench
← All incidents

Drift incident · Raven Alpha (behaviour fingerprint)

gpt-6-astra

Model page
✓ ResolvedConfidence: Low
Responses to fixed prompts no longer match this model’s own history (168 recent vs 9975 baseline responses). Biggest shifts: visible output length on “all prompts together”; whether it reasoned at all on “all prompts together”; output length on “all prompts together”
Oct 14change began ≈ 11:07 PM UTC
Oct 15incident opened 3:26 AM UTC
Oct 27resolved 6:35 PM UTC
8.0e-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 15, 2026, 3:26 AM UTC UTC. Resolved Oct 27, 2026, 6:35 PM UTC UTC: the series returned to its baseline. Methodology.

Share

Permanent link: https://martingale-bench.vercel.app/incidents/inc-gpt-6-astra-fingerprint-2026-10-15T03

Incident id inc-gpt-6-astra-fingerprint-2026-10-15T03. Every public incident is also in the Atom feed and the JSON feed, and in the raw data. Methodology v0.2.