Research
- The Agent Failure Index Published 21 August 2026 The model was at fault in 3 of 358 failures (0.8%).
- The AI Evidence Index Published 18 August 2026 17 of the 21 findings Advizr itself had cited did not match their source.
- A decision model against a chat model for agent routing Internal study The Jev decision model scored 52 of 52 (lower bound 0.931, p95 389 ms). gpt-4o scored 48 of 52 (p95 2,222 ms), and 1 injection got through.
- Per-call context demand and the routing veto Internal study The median call sent 30,596 tokens. A 200K context window holds 99.29% of calls.
Pick a study. It opens here, next to the list.