RKB / LIVE EVALUATION FEEDUPDATING
4,201 concurrent aura observations
Public research release · August 2026
RKB-1.0
The world's first benchmark for
Internet Native Intelligence.
A rigorous evaluation of confidence, charisma, and
culturally persistent output across frontier models.
0Datasets
0Annotated hours
∞Aura tokens
0Confidence
Scroll to begin
01 / Performance frontier
The RKB leaderboard.
A comparative study of models attempting
to reproduce the irreducible.
RKB-1.0 / public evaluation suite
| Rank | Model | Organization | RKB score | Confidence | Status |
|---|
Scores are normalized against a human baseline. Higher is more Ravi. Methodology ↗
Live research telemetry
The benchmark is watching.
Streaming observations from the
public internet, approximately.
02 / Reference archive
High-signal evidence.
A growing corpus of culturally
persistent source material.
Podcast / Conviction
Money follows
my brother.
Pause timing: 99.8 percentileMotion / Locomotion
Gucci Walk
fidelity study.
Locked pigeon coefficient: 1.00Parliament / Latency
Pressure-tested
language systems.
Reporter density: six unitsSource reels are embedded from Instagram and remain subject to their availability and platform controls.
Cross-domain / Generalization
Friends, but
Internet Native.
Transfer learning: unusually strong02 / Evaluation suite
Six tasks. One impossible standard.
Designed to expose the limits of
conventional intelligence.
RKB personal evaluation
Evaluate your
aura.
Submit an image. Our proprietary multimodal system will assess your Internet Native Intelligence across 42 undisclosed dimensions.
⊕ Privacy-preserving⊕ No data retained
RKB / PERSONALv1.0
Upload a photo, GIF, or short videoPNG, JPG, MP4 up to 20 MB
or
03 / Research results
Scaling, but make it iconic.
Our findings indicate that scale helps.
It is not, however, sufficient.
Aura scaling law
RKB score by parameter countLarger models approach—but never reach—Ravi Kishan.
Meme persistence
Exposure decay across 30 days Ravi exposure Conventional meme
The curve does not decay. It develops new properties.
04 / Methodology
Confidence without
comprehension is
measurable.
RKB evaluates models across a multi-modal corpus of public appearances, interviews, parliamentary interventions, and high-signal social artifacts.
RKB = αG + βM + γC + δA
G Gucci WalkM Money ConvictionC ChaosA Aura
Read the technical methodology →Research publication / 2026
Evaluating Internet
Native Intelligence.
The RKB-1.0 technical report, including the full evaluation protocol, scaling observations, and known failure cases.
RKBEvaluating
Ravi Kishan Research Collective
August 202601
RKB-1.0
Evaluating
Internet Native
Intelligence
August 202601