The credit layer for on-chain finance. For every address that trades.

Three sub-models compose one credit model. It returns one score per address — and the whole platform runs on that score.

Coverage

Every address on Hyperliquid, scored.

A complete scan of the venue, refreshed daily — no sampling, no shortlist.

— addresses score above 600.— sit below 300.

Scanned Addresses on Hyperliquid

—

—
above 600
—
300 – 600
—
below 300

Architecture

How a score is made.

Three sub-models read the same address independently. A composite model turns the three readings into one score.

300+
on-chain features read
One score per address · 0–1000 · CC – AAA
How to read a score →
Gravity Index
Trading capability
Social Credit
Network reputation
Eco Engagement
On-chain standing

Applications

What we've built.

Some you can open right now, some still in testing, the Credit API still being built. All of it sits on the same credit engine.

Copy Trading

Only addresses above 600 can be followed. Anyone can follow them.

Open ↗

Earn

Where the highest-scoring addresses agree on a position, an AI review layer decides whether to act — and the quant stack executes.

Open ↗

Trading Terminal

Open to anyone. Orders route to the connected venue — currently Hyperliquid only.

Open ↗

AI trading, step one

HTU Labs

Five research tools for token and market risk. A separate product — sign in with X, each query uses Fuel.

Open ↗

Under-Collateralized Lending

Beta

Borrow against a score instead of an over-collateralized position.

Trader Certification

Beta

Distinguish real trading skill from performative success.

HyperTrend Trading University

Beta

Benchmark a trading agent, then certify what it can actually do.

Credit API

Building

Ask the model your own question — sybil filters, risk gates, address screening.

Status is the current state, not a delivery date

All applications →