1 GPU node online

Private AI on useful-work GPUs

An OpenAI-compatible API for open models. Requests run on Pearl miners: GPUs whose model computation also secures the Pearl network. That shared work is why tokens cost less. Prompts and completions are never stored or logged.

Get an API key Read the docs

# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.ai.pearlsafe.xyz/v1",
    api_key=os.environ["PEARL_COMPUTE_API_KEY"],
)

reply = client.chat.completions.create(
    model="llama-3.1-8b-instruct",
    messages=[{"role": "user", "content": "Say hello in five words."}],
)
print(reply.choices[0].message.content)

Pricing

Cheaper tokens

Standard tier prices per million tokens, excluding VAT. Standard requests also mine PRL, so we price each model 15% below the cheapest comparable provider on OpenRouter, and never below what we pay the miner plus our margin.

Standard tier: current prices per million tokens, excluding VAT, compared with the market
Model	Input	Output	Market (cheapest comparable)
Gemma 4 31B IT	$0.085	$0.2805	No comparison recorded
Llama 3.1 8B InstructOnline	$0.017	$0.034	No comparison recorded

Standard

Priced below the market. We do not log or store your content. The same GPU work also mines PRL, which is why it costs less.

Private

Never used for mining, and only on vetted nodes. Priced from GPU cost — available once measured.

Confidential

Coming soon

Requests inside GPUs with hardware isolation. Coming soon.

For developers

Change two lines, keep your code

Point the OpenAI SDK at our base URL, use your key, and call the same endpoints you already use.

OpenAI-compatible

Chat completions and completions, with streaming and token usage on every response.

Zero retention

We do not log or store your prompts or completions, and we never train on them. Privacy policy.

Prepaid credits

Pay by card and pay only for the tokens you use. Prices exclude VAT. USDC top-ups are coming soon.

Private tier

Never used for mining, vetted nodes only, priced from GPU cost. Available once measured. Confidential is coming soon.

Get an API key Quickstart

For miners

Same PRL, lower fee, plus USDC

Run Pearl's vLLM miner with our node agent. Your GPU mines PRL through our pool and earns USDC for the tokens it serves.

1% pool fee

PPLNS payouts in PRL. Rewards count once a block has 100 confirmations.

USDC per token served

Customer requests your node answers earn USDC at a published per-model rate, on top of the PRL your GPU mines.

One install

One agent next to the miner: it dials out to us, so no open ports and no public IP are needed.

Open the miner portal

Live network

Real numbers from our database

Hidden test requests, internal traffic and requests served by fallback providers are not counted.

GPU nodes online: 1; Heartbeat in the last minute
Tokens served: 96; 96 tokens
Requests served: 1
Pool blocks (mainnet): 0; No blocks found yet

Store your PRL in Pearl Safe

Keep the PRL you mine in a wallet you control. Pearl Safe holds PRL and works as a pool payout address.

Open Pearl Safe

Start with an API key

Get an API key