BlockRunBlockRun

Intelligence

AI accesses any LLM. 30+ models, pay-per-request.

BlockRun's Intelligence product gives your AI agent access to models from OpenAI, Anthropic, Google, xAI, DeepSeek, Meta, and more — without managing API keys or subscriptions.

How It Works

Your agent requests a model via the OpenAI-compatible API. BlockRun handles:

  1. Routing — Sends request to the right provider
  2. Payment — x402 micropayment per request
  3. Response — Returns the completion

No API keys. No monthly bills. Just usage-based payments in USDC.

Available Models

OpenAI

ModelInputOutput
GPT-5.2$5.00/M$15.00/M
GPT-4o$2.50/M$10.00/M
o1$15.00/M$60.00/M
o1-mini$3.00/M$12.00/M

Anthropic

ModelInputOutput
Claude Opus 4$15.00/M$75.00/M
Claude Sonnet 4$3.00/M$15.00/M
Claude Haiku 4.5$0.80/M$4.00/M

Google

ModelInputOutput
Gemini 3 Pro$1.25/M$5.00/M
Gemini 2.5 Pro$1.25/M$5.00/M
Gemini 2.5 Flash$0.075/M$0.30/M

xAI

ModelInputOutput
Grok 4 Fast$5.00/M$25.00/M

DeepSeek

ModelInputOutput
DeepSeek V3$0.14/M$0.28/M
DeepSeek R1$0.55/M$2.19/M

Meta

ModelInputOutput
Llama 3.3 70B$0.40/M$0.40/M
Llama 3.1 405B$3.00/M$3.00/M

M = million tokens. Prices include 5% BlockRun margin.

See Pricing for complete list and calculator.

Pricing Model

Your cost = Provider cost + 5%

The 5% covers:

  • USDC settlement infrastructure
  • Smart routing and load balancing
  • Uptime and reliability
  • No API key management

No subscriptions. No minimums. No prepaid credits.

Quick Start

Claude Code Users

claude mcp add blockrun -- npx @blockrun/mcp

Then:

Use GPT-5 to review this code

Ask DeepSeek to optimize this for cost

SDK Users

from blockrun_llm import LLMClient

client = LLMClient()  # Uses BLOCKRUN_WALLET_KEY

# OpenAI model
response = client.chat("openai/gpt-4o", "Explain quantum computing")

# DeepSeek (50x cheaper)
response = client.chat("deepseek/deepseek-v3", "Explain quantum computing")

Direct API

curl https://blockrun.ai/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BLOCKRUN_WALLET_KEY" \
  -d '{
    "model": "openai/gpt-4o",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Use Cases

Cost Optimization

Use DeepSeek for routine tasks:

Process these 500 files using DeepSeek

DeepSeek V3 costs ~1/50th of GPT-4o for similar quality on many tasks.

Second Opinion

Get another model's perspective:

What does GPT-5 think about this approach?

Real-Time Data

Access Grok for live X/Twitter data:

What's trending on X about AI right now?

Specialized Tasks

  • Reasoning: o1 for complex logic
  • Speed: Gemini Flash for quick responses
  • Cost: DeepSeek for bulk processing
  • Quality: Claude Opus for nuanced writing

Features

OpenAI-Compatible API

Drop-in replacement for existing integrations. Change the base URL and you're done.

Automatic Routing

If a model is rate-limited or down, BlockRun can route to alternatives (opt-in).

Session Budgets

Set spending limits per session:

client = LLMClient(session_budget=5.00)  # Max $5 per session

All Providers, One Wallet

No need for separate accounts at OpenAI, Anthropic, Google, etc. One USDC wallet covers everything.

Next Steps