Catalogue No. 07 · Autumn 2026
Perception research laboratory 図版 01 / 液体レンズ

PRISMA

Fig. 01A model, rendered as glass.

PRISMA Labs builds perception models that take in images, sound and text the way a prism takes in light: everything passes through, every part is separated, nothing is lost.

Chapter I

Three properties of light, three ways of seeing.

Refraction

Pl. I

Vision models that bend toward detail. A hairline crack in a turbine blade, a shadow on an X‑ray, a single misprinted digit on an invoice.

Dispersion

Pl. II

Language models that split a dense document into its parts: claims, numbers, obligations, tone. Each part is shown on its own line, with the source beside it.

Interference

Pl. III

Multimodal fusion. When a camera, a microphone and a report disagree, PRISMA shows where the waves meet and where they cancel out.

Chapter II

Index of models

ModelSeesBandContextLatencyStatus
Clear 2Images, video4,096 frames38 msAvailable
LuxDocuments, text2,000,000 tokens120 msAvailable
HaloSound, speech9 h audio64 msAvailable
SpectrumAll of the aboveMixed210 msPreview
Chapter III · 測定

Measured, not claimed.

Fine detail recognition on the Lumen 40K benchmark, 2026 edition. Higher is better. Every run is published with its settings, so anyone can repeat it.

Sample figures for this concept site

PRISMA Clear 2
91.4
Model A
84.0
Model B
79.6
Model C
71.2
0255075100

Light can bend without being lost.

Write to
the lab.