Notes, benchmarks, and behind-the-scenes posts from the Discera team. No marketing fluff — just what we measured and what we learned.
November 5, 2026
Our biggest router test yet: 1,000 completely new questions — none from any earlier test — each run twice. 98.5% on the questions the router can answer, 92.0% across all 1,000, and both numbers explained.
Read more →
October 26, 2026
The rebuilt 1.5 router, graded harder than 1.0 was: the original 147 questions plus two completely new sets written after the rebuild. It averaged 94.1% — against the 80% target we set in public.
Read more →
October 19, 2026
There’s no single best AI — it depends on the task. Writing, coding, math, research, and images, broken down honestly — and the case for never having to choose.
Read more →
October 10, 2026
The router is the entire product — one chat that picks the best AI for each message. A claim like that needs numbers, so we publish ours: the test, the scores, and where it’s wrong.
Read more →