BlockRun LLM SDK

Pay-per-request access to GPT-5.2, Claude 4, Gemini 2.5, Grok, and more via x402 micropayments.

BlockRun assumes Claude Code as the agent runtime.

Supported Chains

Chain	Network	Payment	Status
Base	Base Mainnet (Chain ID: 8453)	USDC	✅ Primary
Base Testnet	Base Sepolia (Chain ID: 84532)	Testnet USDC	✅ Development
XRPL	XRP Ledger Mainnet	RLUSD	✅ New

Protocol: x402 v2

Installation

pip install blockrun-llm

Quick Start

from blockrun_llm import LLMClient

client = LLMClient()  # Uses BLOCKRUN_WALLET_KEY (never sent to server)
response = client.chat("openai/gpt-5.2", "Hello!")

That's it. The SDK handles x402 payment automatically.

Smart Routing (ClawRouter)

Let the SDK automatically pick the cheapest capable model for each request:

from blockrun_llm import LLMClient

client = LLMClient()

# Auto-routes to cheapest capable model
result = client.smart_chat("What is 2+2?")
print(result.response)  # '4'
print(result.model)     # 'nvidia/kimi-k2.5' (cheap, fast)
print(f"Saved {result.routing.savings * 100:.0f}%")  # 'Saved 94%'

# Complex reasoning task -> routes to reasoning model
result = client.smart_chat("Prove the Riemann hypothesis step by step")
print(result.model)  # 'xai/grok-4-1-fast-reasoning'

Routing Profiles

Profile	Description	Best For
free	nvidia/gpt-oss-120b only (FREE)	Testing, development
eco	Cheapest models per tier (DeepSeek, xAI)	Cost-sensitive production
auto	Best balance of cost/quality (default)	General use
premium	Top-tier models (OpenAI, Anthropic)	Quality-critical tasks

# Use premium models for complex tasks
result = client.smart_chat(
    "Write production-grade async Python code",
    routing_profile="premium"
)
print(result.model)  # 'anthropic/claude-opus-4.5'

How It Works

ClawRouter uses a 14-dimension rule-based classifier to analyze each request:

• Token count - Short vs long prompts
• Code presence - Programming keywords
• Reasoning markers - "prove", "step by step", etc.
• Technical terms - Architecture, optimization, etc.
• Creative markers - Story, poem, brainstorm, etc.
• Agentic patterns - Multi-step, tool use indicators

The classifier runs in <1ms, 100% locally, and routes to one of four tiers:

Tier	Example Tasks	Auto Profile Model
SIMPLE	"What is 2+2?", definitions	nvidia/kimi-k2.5
MEDIUM	Code snippets, explanations	xai/grok-code-fast-1
COMPLEX	Architecture, long documents	google/gemini-3-pro-preview
REASONING	Proofs, multi-step reasoning	xai/grok-4-1-fast-reasoning

How It Works

1. You send a request to BlockRun's API
2. The API returns a 402 Payment Required with the price
3. The SDK automatically signs a USDC payment on Base
4. The request is retried with the payment proof
5. You receive the AI response

Your private key never leaves your machine - it's only used for local signing.

Available Models

OpenAI GPT-5 Family

Model	Input Price	Output Price
openai/gpt-5.2	$1.75/M	$14.00/M
openai/gpt-5-mini	$0.25/M	$2.00/M
openai/gpt-5-nano	$0.05/M	$0.40/M
openai/gpt-5.2-pro	$21.00/M	$168.00/M

OpenAI GPT-4 Family

Model	Input Price	Output Price
openai/gpt-4.1	$2.00/M	$8.00/M
openai/gpt-4.1-mini	$0.40/M	$1.60/M
openai/gpt-4.1-nano	$0.10/M	$0.40/M
openai/gpt-4o	$2.50/M	$10.00/M
openai/gpt-4o-mini	$0.15/M	$0.60/M

OpenAI O-Series (Reasoning)

Model	Input Price	Output Price
openai/o1	$15.00/M	$60.00/M
openai/o1-mini	$1.10/M	$4.40/M
openai/o3	$2.00/M	$8.00/M
openai/o3-mini	$1.10/M	$4.40/M
openai/o4-mini	$1.10/M	$4.40/M

Testnet Models (Base Sepolia)

Model	Price
openai/gpt-oss-20b	$0.001/request
openai/gpt-oss-120b	$0.002/request

Testnet models use flat pricing (no token counting) for simplicity.

Anthropic Claude

Model	Input Price	Output Price
anthropic/claude-opus-4.5	$5.00/M	$25.00/M
anthropic/claude-opus-4	$15.00/M	$75.00/M
anthropic/claude-sonnet-4	$3.00/M	$15.00/M
anthropic/claude-haiku-4.5	$1.00/M	$5.00/M

Google Gemini

Model	Input Price	Output Price
google/gemini-3-pro-preview	$2.00/M	$12.00/M
google/gemini-2.5-pro	$1.25/M	$10.00/M
google/gemini-2.5-flash	$0.15/M	$0.60/M

DeepSeek

Model	Input Price	Output Price
deepseek/deepseek-chat	$0.28/M	$0.42/M
deepseek/deepseek-reasoner	$0.28/M	$0.42/M

xAI Grok

Model	Input Price	Output Price	Context	Notes
xai/grok-3	$3.00/M	$15.00/M	131K	Flagship
xai/grok-3-fast	$5.00/M	$25.00/M	131K	Tool calling optimized
xai/grok-3-mini	$0.30/M	$0.50/M	131K	Fast & affordable
xai/grok-4-1-fast-reasoning	$0.20/M	$0.50/M	2M	Latest, chain-of-thought
xai/grok-4-1-fast-non-reasoning	$0.20/M	$0.50/M	2M	Latest, direct response
xai/grok-4-fast-reasoning	$0.20/M	$0.50/M	2M	Step-by-step reasoning
xai/grok-4-fast-non-reasoning	$0.20/M	$0.50/M	2M	Quick responses
xai/grok-code-fast-1	$0.20/M	$1.50/M	256K	Code generation
xai/grok-4-0709	$0.20/M	$1.50/M	256K	Premium quality
xai/grok-2-vision	$2.00/M	$10.00/M	32K	Vision capabilities

Moonshot Kimi

Model	Input Price	Output Price
moonshot/kimi-k2.5	$0.50/M	$2.40/M

NVIDIA (Free & Hosted)

Model	Input Price	Output Price	Notes
nvidia/gpt-oss-120b	FREE	FREE	OpenAI open-weight 120B (Apache 2.0)
nvidia/kimi-k2.5	$0.55/M	$2.50/M	Moonshot 1T MoE with vision

E2E Verified Models

All models below have been tested end-to-end via the Python SDK (Feb 2026):

Provider	Model	Status
OpenAI	openai/gpt-4o-mini	Passed
Anthropic	anthropic/claude-sonnet-4	Passed
Google	google/gemini-2.5-flash	Passed
DeepSeek	deepseek/deepseek-chat	Passed
xAI	xai/grok-3-fast	Passed
Moonshot	moonshot/kimi-k2.5	Passed

Image Generation

Model	Price
openai/dall-e-3	$0.04-0.08/image
openai/gpt-image-1	$0.02-0.04/image
black-forest/flux-1.1-pro	$0.04/image
google/nano-banana	$0.05/image
google/nano-banana-pro	$0.10-0.15/image

Usage Examples

Simple Chat

from blockrun_llm import LLMClient

client = LLMClient()  # Uses BLOCKRUN_WALLET_KEY (never sent to server)

response = client.chat("openai/gpt-5.2", "Explain quantum computing")
print(response)

# With system prompt
response = client.chat(
    "anthropic/claude-sonnet-4",
    "Write a haiku",
    system="You are a creative poet."
)

Real-time X/Twitter Search (xAI Live Search)

Note: Live Search can take 30-120+ seconds as it searches multiple sources. The SDK automatically uses a 5-minute timeout for search requests.

from blockrun_llm import LLMClient

client = LLMClient()

# Simple: Enable live search with search=True (default 10 sources, ~$0.26)
response = client.chat(
    "xai/grok-3",
    "What are the latest posts from @blockrunai?",
    search=True
)
print(response)

# Custom: Limit sources to reduce cost (5 sources, ~$0.13)
response = client.chat(
    "xai/grok-3",
    "What's trending on X?",
    search_parameters={"mode": "on", "max_search_results": 5}
)

# Custom timeout (if 5 min isn't enough)
client = LLMClient(search_timeout=600.0)  # 10 minutes

Check Spending

from blockrun_llm import LLMClient

client = LLMClient()

response = client.chat("openai/gpt-5.2", "Explain quantum computing")
print(response)

# Check how much was spent
spending = client.get_spending()
print(f"Spent ${spending['total_usd']:.4f} across {spending['calls']} calls")

Full Chat Completion

from blockrun_llm import LLMClient

client = LLMClient()  # Uses BLOCKRUN_WALLET_KEY (never sent to server)

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "How do I read a file in Python?"}
]

result = client.chat_completion("openai/gpt-5.2", messages)
print(result.choices[0].message.content)

Async Usage

import asyncio
from blockrun_llm import AsyncLLMClient

async def main():
    async with AsyncLLMClient() as client:
        # Simple chat
        response = await client.chat("openai/gpt-5.2", "Hello!")
        print(response)

        # Multiple requests concurrently
        tasks = [
            client.chat("openai/gpt-5.2", "What is 2+2?"),
            client.chat("anthropic/claude-sonnet-4", "What is 3+3?"),
            client.chat("google/gemini-2.5-flash", "What is 4+4?"),
        ]
        responses = await asyncio.gather(*tasks)
        for r in responses:
            print(r)

asyncio.run(main())

List Available Models

from blockrun_llm import LLMClient

client = LLMClient()
models = client.list_models()

for model in models:
    print(f"{model['id']}: ${model['inputPrice']}/M input, ${model['outputPrice']}/M output")

Testnet Usage

For development and testing without real USDC, use the testnet:

from blockrun_llm import testnet_client

# Create testnet client (uses Base Sepolia)
client = testnet_client()  # Uses BLOCKRUN_WALLET_KEY

# Chat with testnet model
response = client.chat("openai/gpt-oss-20b", "Hello!")
print(response)

# Check testnet USDC balance
balance = client.get_balance()
print(f"Testnet USDC: ${balance:.4f}")

Testnet Setup

1. Get testnet ETH from Alchemy Base Sepolia Faucet
2. Get testnet USDC from Circle USDC Faucet
3. Set your wallet key: export BLOCKRUN_WALLET_KEY=0x...

Available Testnet Models

• openai/gpt-oss-20b - $0.001/request (flat price)
• openai/gpt-oss-120b - $0.002/request (flat price)

Manual Testnet Configuration

from blockrun_llm import LLMClient

# Or configure manually
client = LLMClient(api_url="https://testnet.blockrun.ai/api")
response = client.chat("openai/gpt-oss-20b", "Hello!")

XRPL Chain (RLUSD Payments)

BlockRun now supports payments with RLUSD on the XRP Ledger. Same models, same API - just a different payment rail.

from blockrun_llm import xrpl_client

# Create XRPL client (pays with RLUSD)
client = xrpl_client()  # Uses BLOCKRUN_WALLET_KEY

# Chat with any model
response = client.chat("openai/gpt-4o", "Hello!")
print(response)

# Check RLUSD balance
balance = client.get_balance()
print(f"RLUSD: ${balance:.4f}")

Async XRPL Usage

import asyncio
from blockrun_llm import async_xrpl_client

async def main():
    async with async_xrpl_client() as client:
        response = await client.chat("openai/gpt-4o", "Hello!")
        print(response)

asyncio.run(main())

Manual XRPL Configuration

from blockrun_llm import LLMClient

# Or configure manually
client = LLMClient(api_url="https://xrpl.blockrun.ai/api")
response = client.chat("openai/gpt-4o", "Hello!")

Environment Variables

Variable	Description	Required
BLOCKRUN_WALLET_KEY	Your Base chain wallet private key	Yes (or pass to constructor)
BLOCKRUN_API_URL	API endpoint	No (default: https://blockrun.ai/api)

Setting Up Your Wallet

1. Create a wallet on Base network (Coinbase Wallet, MetaMask, etc.)
2. Get some ETH on Base for gas (small amount, ~$1)
3. Get USDC on Base for API payments
4. Export your private key and set it as BLOCKRUN_WALLET_KEY

# .env file
BLOCKRUN_WALLET_KEY=0x...your_private_key_here

Error Handling

from blockrun_llm import LLMClient, APIError, PaymentError

client = LLMClient()

try:
    response = client.chat("openai/gpt-5.2", "Hello!")
except PaymentError as e:
    print(f"Payment failed: {e}")
    # Check your USDC balance
except APIError as e:
    print(f"API error ({e.status_code}): {e}")

Testing

Running Unit Tests

Unit tests do not require API access or funded wallets:

pytest tests/unit                    # Run unit tests only
pytest tests/unit --cov              # Run with coverage report
pytest tests/unit -v                 # Verbose output

Running Integration Tests

Integration tests call the production API and require:

• A funded Base wallet with USDC ($1+ recommended)
• BLOCKRUN_WALLET_KEY environment variable set
• Estimated cost: ~$0.05 per test run

export BLOCKRUN_WALLET_KEY=0x...
pytest tests/integration             # Run integration tests only
pytest                               # Run all tests

Integration tests are automatically skipped if BLOCKRUN_WALLET_KEY is not set.

Security

Private Key Safety

• Private key stays local: Your key is only used for signing on your machine
• No custody: BlockRun never holds your funds
• Verify transactions: All payments are on-chain and verifiable

Best Practices

Private Key Management:

• Use environment variables, never hard-code keys
• Use dedicated wallets for API payments (separate from main holdings)
• Set spending limits by only funding payment wallets with small amounts
• Never commit .env files to version control
• Rotate keys periodically

Input Validation: The SDK validates all inputs before API requests:

• Private keys (format, length, valid hex)
• API URLs (HTTPS required for production, HTTP allowed for localhost)
• Model names and parameters (ranges for max_tokens, temperature, top_p)

Error Sanitization: API errors are automatically sanitized to prevent sensitive information leaks.

Monitoring:

address = client.get_wallet_address()
print(f"View transactions: https://basescan.org/address/{address}")

Keep Updated:

pip install --upgrade blockrun-llm  # Get security patches

Links

• Website
• Documentation
• GitHub
• Discord

License

MIT