The Surprise Intelligence Model
Confidence,
checked.
Zeno checks predictions from models you already use. It shows where they may fail, an adjusted forecast, and the evidence.
Frozen example · 24 hours ahead
Sydney ensemble forecast
panel
26.5°
Zeno
25.1°
actual
22.2°
01 · evidence
Later weeks. No peeking.
Zeno improved five named sources. Where it lost, the source reference is kept.
02 · warning
Who breaks first?
The warning read from Zeno's forecast beat a tuned baseline on ensemble weather.
03 · sources
One network. Different evidence.
The weights are shared. The permission to correct is earned source by source.
Crypto price ranges
Zeno correction
+11.7%
Ensemble weather
Zeno correction
+5.6%
Deterministic weather
correction · no history
+6.2%
COVID-19 hospitalisations
correction · no history
+11.8%
Sea conditions
correction · no history
+3.2%
Prediction markets
reference kept
RSV
reference kept
FluSight
reference kept
ForecastBench
reference kept
Numinous AI forecasters
learn-only
04 · curriculum
Crowds leave fingerprints.
323 AI forecasters, arranged by how similarly they forecast and linked where they missed together.
05 · one answer
From question to receipt.
Zeno30 gathers inputs, Offdiagonal verifies evidence, Zeno assesses reliability, Arena records performance. API and MCP deliver the same answer.
Supporting system
Arena · Offdiagonal · NVIDIA Inception · $ZENO