Nano-prompt runtime · v0.9

Fold a model into a single filament.

LiLo compresses nano-prompts, webMCP tooling, and AI Digital Twin workflows into models up to 100× lighter — so full intelligence runs on the smallest edge.

100×
lighter
1.4k
params
9.2k
tok/s
132
ms p50
lilo · compressACTIVE

$ lilo run --twin --edge

→ nano-prompt folded · 1.4k params

→ webMCP bound · 41 tools

→ digital twin online · 132ms p50

✓ 100× lighter · 9.2k tok/s

model

lilo-nano

twin

online

tools

41

Capabilities

One dense filament, three folded systems.

01

Nano-prompts

Prompts distilled to their load-bearing tokens, so a full intent ships in under 40 bytes.

02

webMCP

A model-control plane that binds 41+ web tools as a single compact control surface.

03

AI Digital Twin

A live, lightweight replica of your workflow that reasons at the edge without a datacenter.

Efficiency

100× lighter, engineered to the byte.

We measure the whole stack — params, tokens per second, time-to-first-token. LiLo holds the line on all three, on a 2-core edge box.

100×vs. baseline
frontier
metriclilo-nano / frontier
Parameters1.4k vs 70B
Tokens / sec9.2k vs 940
Time-to-first-token132ms vs 4.1s
Memory footprint38MB vs 14GB

Bench · 2-core edge · 8GB RAM · 2026-Q1

Pricing

Priced per filament, not per token.

Starter

For a single workflow.

$0 /mo

  • 1 nano-prompt template
  • 10 webMCP tools
  • 1k twin runs / mo
Start free
Most chosen

Pro

For a lean product team.

$49 /mo

  • Unlimited nano-prompts
  • 41 webMCP tools
  • 50k twin runs / mo
  • Edge + regional deploy
Join Pro waitlist

Team

For fleets of twins.

$199 /mo

  • Everything in Pro
  • Unlimited twin runs
  • SSO + audit log
Talk to us

Waitlist

Reserve your filament.

Early access opens in cohorts. Leave your details and we'll fold you into the next build of LiLo by KAADE CONSTRUCTION.

Cohort 03 · opens 2026-Q2

Priority to edge-only teams

No spam · Cancel anytime