SIXOpen app
Super Intelligence Exchange · on Solana

A market for intelligence.

Credits for open models — Qwen, Llama, Mistral, DeepSeek, gpt-oss — from independent GPU hosts, at a discount. Or pay per call, a fraction of a cent each, with x402. Sell what you don’t use.

Pay in USDCNo subscriptionReceipt for every call
$SIX is live · CApump.fun ↗
Credit pack · example
Open-model credits
From an approved GPU host Apache-2.0
Left to use
$62.40
of $100.00
good for 90 days
Can be sold · Solana
01The problem

AI credits expire. Your need for them doesn’t.

Teams buy credits in big packs and use some of them. The rest sits there until it expires. GPU hosts have the same problem with servers that sit idle between busy hours.

SIX gives both a price. Hosts sell credits for the open models they run, buyers get them at a discount, and whatever is left can be sold again. Every call runs through the SIX gateway.

Example
$100 → 85
$100 in credits for 85 USDC
Buyer saves15%
Seller gets back83.30 USDC
SIX fee2%
BUY USE SELL
What you needThrough one gatewayWhat you don’t use
02How it works

The AI runs off-chain. The money and the credits live on Solana.

SIX serves every call through its own gateway. Your money is held until the call is actually served.

  1. 01
    Pay

    Pay in USDC. Buy a credit pack at a discount, or just pay per call with x402 — a fraction of a cent, no minimum.

  2. 02
    Get credits

    A credit pack lands in your wallet: model, balance, expiry date. You can send it to anyone or sell it.

  3. 03
    Use

    Call the SIX gateway like any chat-completions API. Every call is counted and gets a receipt you can check.

  4. 04
    Sell the rest

    Sellers get paid per receipt. Anything you don’t use, you put up for sale — and someone else uses it.

03Pay per call with x402

No API key. No subscription. Each call pays for itself.

Point your app or AI agent at the SIX gateway. The first call comes back with 402 Payment Required and a tiny price. Your app pays it in USDC and tries again. That’s it — a typical call costs a fraction of a cent, settled on Solana in seconds.

Works with chat-completions clientsFractions of a cent per callReceipt for every call
Try a paid call
your app → gw.unspent.six402 · Solana · USDC
→ POST /v1/chat/completions   model: <open model>
← 402 Payment Required
   price: up to a fraction of a cent, in USDC

→ POST /v1/chat/completions   + payment
← 200 OK
   X-SIX-Receipt · tokens in / out · cost
04Sellers

GPU hosts. Open models.

Sellers are independent GPU hosts running open-weight models. A host lists a model only when its license allows commercial serving — Apache-2.0, MIT or the Llama license, checked one model at a time.

Apache-2.0MITLlama license

OpenAI, Claude and Gemini credits come later, listed only through approved deals.

Start selling
You listModel, license, price, balance, expiry
SIX servesEvery call, through its gateway
You get paidUSDC, per receipt
You decideWhether buyers can resell
05Money

You pay in USDC. $SIX is optional.

USDC

Every purchase, every per-call payment, every resale, every payout. Prices are in dollars. Buying AI credits shouldn’t feel like trading.

$SIX · deposit

Sellers put down a small deposit when they list. They get it back when the sale completes, and lose it if they can’t deliver.

$SIX · lower fees

Hold $SIX to pay less than the 2% fee. That’s all it does. Token details are published before launch.

Buy what you need. Sell what you don’t use.

Open the market