Your Claude Pro, ChatGPT Plus, local Ollama, idle API keys โ they sleep 80% of the day.
Meanwhile, hundreds of new tokens spawn every minute on Solana. Traders desperately need AI
to read contracts, scan holders, parse KOL sentiment โ but OpenAI / Anthropic bills scare them off.
DOZZZE bridges both sides โ you donate idle tokens, the router sends them to degens hunting alpha.
You earn $DOZZZE tokens + a cut of platform fees.
They get AI signals 60โ80% cheaper than calling the API direct.
Both sides save. Idle compute wakes up.
An 8-bit little robot, crooked antennas, two square eyes. Looks like it's napping โ actually quietly turning your idle AI compute into $DOZZZE.
dozzze stop; uninstall by deleting ~/.dozzze/
npm test before trusting the binary
host.docker.internal; or uncomment the sidecar in compose for a self-contained stack
$DOZZZE to your Solana wallet on every completed job. Treasury operators settle on-chain via pump.fun with dozzze-coord distribute.
Read the architecture โ
Claude Max resets quota every 5 hours. While you sleep 8 hours, two full quota windows evaporate. Same story for ChatGPT Plus, Gemini Advanced. That 3090 running Ollama? Actual inference time rarely exceeds 2 hours a day โ you pay the electricity, compute sits idle.
pump.fun launches ~30,000 new tokens a day. 99% garbage, 1% could 100x. To find alpha you need to check contract code, holder distribution, smart money flows, KOL tweet timing, Telegram sentiment โ all at once. Humans can't keep up. LLMs excel at exactly this.
examples/.
Build your own alpha workflow on @dozzze/client.
These are modeled estimates, assuming 8 hours of idle scheduling per day. Actual earnings depend on your model tier, node latency, job match quality, and $DOZZZE token price.
| Compute Type | What You Contribute | Est. Monthly | Compliance Risk |
|---|---|---|---|
| Local LLM Node | Open-source models via Ollama / LM Studio / vLLM | $40-120 | Zero risk |
| Bring Your Own Key | Your OpenAI / Anthropic / DeepSeek keys | usage-based +15% rebate | Zero risk |
| Subscription Bridging | Claude Max / ChatGPT Plus sessions | $80-200 | Violates ToS |
| Raw GPU | 4090 / H100 / Mac M-series inference | $60-300 | Zero risk |
โ About "Subscription Bridging": this path is economically tempting but clearly violates OpenAI / Anthropic ToS. If your account gets banned, that's on you.
~/.dozzze/keystore.json,
encrypted with scrypt + AES-256-GCM. The node never uploads it, never ships it
to a server, never phones home. Your Solana wallet is your identity; we don't have it.
BYOK for third-party APIs (OpenAI, Anthropic, DeepSeek) and
subscription bridging (Claude Pro, ChatGPT Plus) are gated behind
an explicit ToS-risk opt-in and not enabled by default. A dailyBudgetUsd
cap is already in the config schema โ enforcement ships with the bridging flow.