Intelligence Pricing

Pay only for what you use. Per-token chat is billed at provider cost — no platform margin.

Pricing Formula

Your cost = Provider cost          (per-token chat — no platform margin)
              + $0.001 / request     (flat x402 transaction fee)

Media generation and Live Search still carry a 5% platform margin, which covers:

• x402 settlement infrastructure
• Smart routing and reliability
• No API key management
• Instant on-chain payments

What $1 Gets You

Full Price List

Cached input is the rate for prompt tokens the provider serves from its prompt cache — a repeated system prompt, a long conversation prefix. The provider decides what is a cache hit and reports it in the response usage (prompt_tokens_details.cached_tokens). A dash means the provider publishes no separate cached rate, so cached tokens bill at the input rate.

How the cached rate reaches an x402 payment depends on the scheme the client signs:

• upto — the client signs a ceiling and is charged actual usage after the call. On DeepSeek models, cached tokens are charged at the cached rate; other models bill cached tokens at the input rate. Offered on DeepSeek and Anthropic models, and it needs no ETH (the USDC approval is a gasless signature).
• exact — the client signs the price quoted before the call. Nobody knows about a cache hit yet, so the quote uses the full input rate.

OpenAI

Anthropic

Google

Gemini Pro models double the input rate and add 50% to the output rate above 200K prompt tokens (the whole request reprices), mirroring Google's official long-context pricing — e.g. Gemini 2.5 Pro is $2.50 in · $15.00 out above the threshold. Flash tiers are flat.

xAI Grok

Grok doubles the per-token rates above 200K prompt tokens (the whole request reprices — e.g. Grok 4.7, 4.6 and 4.5 are $4.00 in · $12.00 out above the threshold), mirroring xAI's official long-context tier. Live Search adds one flat surcharge per call: $0.30 on Grok 4.5 / 4.6 / 4.7 / 4.20, $0.12 on Grok 4.3 and below.

Z.AI

Moonshot

MiniMax

Qwen

Tencent

Poolside

Mistral

DeepSeek

Free Tier (8 models)

The free tier is 8 reasoning, coding, and vision models with no per-token charge. The lineup is kept current by a self-healing health gate that routes around any model whose upstream is temporarily unavailable and auto-recovers it, so existing calls keep working. All free-tier models are FREE for both input and output — call GET /api/v1/models for the current live list.

Image Generation

Other media: video from $0.05/sec, music $0.15/track, text-to-speech $0.05–$0.10 per 1k characters, sound effects $0.0535/generation.

Cost Comparison: BlockRun vs Direct

Chat is billed at the provider's list rate with no margin, plus a flat $0.001 transaction fee per request, and you get:

• No API key management
• No monthly invoices
• No prepaid credits
• One wallet for all providers
• Instant per-request settlement

Budget Management

Session Budgets

from blockrun_llm import LLMClient

# Limit spending per session
client = LLMClient(max_session_cost=5.00)

Check Balance

balance = client.get_balance()
print(f"${balance} USDC remaining")

Track Spending

# Get usage stats
usage = client.get_spending()
print(f"Spent: ${usage['total_usd']}")
print(f"Requests: {usage['calls']}")

Cost Optimization Tips

0. Use ClawRouter for Automatic Savings

Save 84% on average with ClawRouter — it automatically routes each request to the cheapest model that can handle it.

/model blockrun/auto

ClawRouter does all the optimization below automatically.

1. Use Cheaper Models for Routine Tasks

# Expensive
response = client.chat("openai/gpt-5.5", "Summarize this text")

# ~36x cheaper, similar quality
response = client.chat("deepseek/deepseek-chat", "Summarize this text")

2. Use Flash Models for Speed

# For quick, simple tasks
response = client.chat("google/gemini-3.5-flash", prompt)

3. Match Model to Task

4. Optimize Prompts

Shorter prompts = fewer input tokens = lower cost.

What you don't pay

• No subscriptions
• No prepaid credits
• No minimum top-up
• No overage charges
• No rate limit fees

The rates on this page are the providers' own published prices; chat tokens carry no margin — only the flat $0.001 transaction fee per request. Media generation and Live Search carry 5%.

Payment Details

• Currency: USDC on Base or Solana
• Settlement: Instant, on-chain
• Verification: Basescan

What's next?