Wrap the standard openai.OpenAI client with a crypto wallet.
When the server responds with HTTP 402, the library automatically signs and retries the request — zero code changes needed.
pip install x402-openai[evm] # Ethereum / Base / …
pip install x402-openai[svm] # Solana
pip install x402-openai[all] # all chainsfromx402_openaiimportX402OpenAIfromx402_openai.walletsimportEvmWalletclient=X402OpenAI(wallet=EvmWallet(private_key="0x…"))
res=client.chat.completions.create(
model="openai/gpt-4o-mini",
messages=[{"role": "user", "content": "Hello!"}],
)
print(res.choices[0].message.content)Swap EvmWallet for SvmWallet to pay on Solana — the API is identical.
fromx402_openaiimportAsyncX402OpenAIclient=AsyncX402OpenAI(wallet=EvmWallet(private_key="0x…"))
stream=awaitclient.chat.completions.create(
model="openai/gpt-4o-mini",
messages=[{"role": "user", "content": "Explain x402"}],
stream=True,
)
asyncforchunkinstream:
ifchunk.choices[0].delta.content:
print(chunk.choices[0].delta.content, end="")fromx402_openai.walletsimportEvmWallet, SvmWalletclient=X402OpenAI(wallets=[
EvmWallet(private_key="0x…"),
SvmWallet(private_key="base58…"),
])wallet=EvmWallet(mnemonic="word1 word2 … word12")
wallet=EvmWallet(mnemonic="…", account_index=2) # m/44'/60'/0'/0/2wallet=EvmWallet(mnemonic="…", derivation_path="m/44'/60'/2'/0/0") # custom pathThe protocol selects the right chain automatically based on the server's payment requirements.
Use policies to control which chain or scheme is preferred when multiple payment options are available:
fromx402_openaiimportX402OpenAI, prefer_network, prefer_scheme, max_amountfromx402_openai.walletsimportEvmWallet, SvmWalletclient=X402OpenAI(
wallets=[
EvmWallet(private_key="0x…"),
SvmWallet(private_key="base58…"),
],
policies=[
prefer_network("eip155:8453"), # Prefer Base mainnetprefer_scheme("exact"), # Prefer exact payment schememax_amount(1_000_000), # Cap at 1 USDC (6 decimals)
],
)Drop-in replacement for openai.OpenAI / openai.AsyncOpenAI. Provide exactly one credential source:
| Parameter | Type | Description |
|---|---|---|
wallet | Wallet | Single wallet adapter |
wallets | list[Wallet] | Multiple adapters (multi-chain) |
policies | list[Policy] | Payment policies (chain/scheme preference, amount cap) |
x402_client | x402HTTPClient* | Pre-configured x402 client (bypasses policies) |
All standard OpenAI kwargs (base_url, timeout, max_retries, …) are forwarded.
Default base_url: https://llm.qntx.org/v1
| Class | Chain | Extra |
|---|---|---|
EvmWallet(private_key=…) | EVM | x402-openai[evm] |
EvmWallet(mnemonic=…) | EVM (BIP-39) | x402-openai[evm] |
SvmWallet(private_key=…) | Solana | x402-openai[svm] |
Implement the Wallet protocol to add a new chain.
X402Transport / AsyncX402Transport — httpx transports for manual wiring into any httpx.Client.
See the examples/ directory. Each script is self-contained:
EVM_PRIVATE_KEY="0x…" python examples/chat_evm.py
SOLANA_PRIVATE_KEY="base58…" python examples/chat_svm.py
EVM_PRIVATE_KEY="0x…" python examples/streaming_evm.py
MNEMONIC="word1 word2 …" python examples/chat_evm_mnemonic.py
EVM_PRIVATE_KEY="0x…" python examples/chat_evm_policy.py
EVM_PRIVATE_KEY="0x…" python examples/streaming_evm_policy.pyThis project is licensed under the MIT License.
