The 0G Compute Python SDK is the official Python interface to the 0G Compute Network, a decentralized GPU marketplace that provides AI inference services including Large Language Models (LLM), text-to-image generation, speech-to-text, and fine-tuning. Providers list AI services on-chain and users pay per request with 0G tokens. All billing, routing, and verification is handled by smart contracts — no accounts, no API keys, no vendor lock-in.

​
What you can do with the SDK

  • Run AI inference — chat with hosted LLMs, generate images, transcribe speech
  • Fine-tune models — upload datasets, train LoRA adapters, deploy to inference GPUs
  • Verify computation — validate TEE attestations and signed responses on-chain
  • Manage funds — deposit, transfer, and refund prepaid balances per provider
  • Issue API keys — create persistent, revocable tokens for server applications

​
Supported service types

  • Chatbot services — conversational AI with OpenAI-compatible endpoints
  • Text-to-image — generate images from text prompts
  • Image editing — modify existing images
  • Speech-to-text — transcribe audio to text
  • Fine-tuning — train custom LoRA adapters on your own datasets

​
How it works

1

Fund your account

Deposit 0G tokens into the Ledger contract. Minimum 3 OG to create a ledger.
2

Discover providers

Query the Inference contract for available models, pricing, and endpoints.
3

Acknowledge a provider

One-time on-chain action that records the provider’s TEE signer to your account.
4

Make requests

The SDK generates a session token; you send OpenAI-compatible HTTP requests.
5

Settle on-chain

The provider deducts tokens from your sub-account based on actual usage.

​
Why 0G Compute

  • Up to 90% cheaper — pay only for compute used, no monthly minimums. Compare to 5,000–50,000/monthfordedicatedcloudGPUsor5,000–50,000/month for dedicated cloud GPUs or 0.03+ per request on managed APIs.
  • Instantly available — access thousands of GPUs globally with 50–100ms latency
  • OpenAI-compatible — drop-in replacement at the HTTP layer; works with the OpenAI Python SDK
  • TEE-verified — every response is cryptographically signed by a Trusted Execution Environment
  • Decentralized — providers are matched on-chain, no single vendor controls access

​
Next steps

Install the SDK

Set up your environment and create a broker in under 5 minutes.

Run your first inference

Fund an account and make a chat completion request.
For protocol-level details and ecosystem docs, see the official 0G Labs documentation.