blockrun

BlockRun - Claude Code's Multi-Model Extension Capabilities

Skill Overview

BlockRun is an extension skill for Claude Code that, via the x402 micropayment wallet, provides AI image generation, real-time Twitter/X data access, and multi-model invocation capabilities without requiring API keys, on a pay-as-you-go basis.

Applicable Scenarios

1. AI Image Generation

When Claude cannot generate images directly, BlockRun can call DALL-E 3 or Google Nano Banana to complete image generation tasks. Each image costs only $0.01–$0.04, suitable for illustrations, logos, product display images, and similar scenarios.

2. Real-Time Social Media Data Retrieval

When you need the latest updates, trend analysis, or content from specific accounts on Twitter/X, BlockRun can use Grok 3's Live Search to perform real-time searches, supporting filters by likes, views, and other criteria.

3. Multi-Model Comparison and Cost Optimization

When a second opinion is needed for code review, or when processing large volumes of text, BlockRun offers a variety of models from GPT-5.2 to DeepSeek, allowing flexible switching according to task complexity and budget.

Core Features

Wallet Management and Pay-As-You-Go Payments

BlockRun uses a USDC wallet on the Base network for payments. The wallet is created automatically on first use and can be topped up via QR code. All fees are charged based on actual usage, with real-time balance and expense tracking available. You can also set budget limits to automatically stop calls.

Unified Multi-Model Invocation Interface

Supports multiple mainstream models—OpenAI (GPT-5.2, GPT-5-mini, o4-mini, o3, DALL-E 3), xAI (Grok 3), DeepSeek, Google (Gemini 2.5 Flash, Nano Banana), etc.—via a unified Python SDK interface, without needing separate API keys for each platform.

Advanced Search and Image Generation

Grok Live Search supports custom data sources (Web, News, X/Twitter) and precise filtering by time range, website source, account, interaction volume, and more. Image generation offers DALL-E 3 (photo-realistic) and Nano Banana (fast artistic-style generation).

Frequently Asked Questions

How is BlockRun charged, and how much budget should I prepare?

BlockRun charges based on actual usage. Text models are billed per million tokens—for example, GPT-5.2 input $1.75/M, output $14/M; image generation costs $0.01–$0.04 per image; Grok live search is $0.025 per data source. It is recommended to top up $1–$5 USDC initially for testing; actual costs depend on usage frequency and model choices.

How do I top up the BlockRun wallet?

Calling setup_agent_wallet() for the first time will automatically create a wallet and display the top-up address and QR code. Transfer USDC to that wallet address (on the Base network). After topping up, you can check the balance via client.get_balance().

Which AI models does BlockRun support?

Currently supports OpenAI's GPT-5.2, GPT-5-mini, o4-mini, o3, and DALL-E 3; xAI's Grok 3; DeepSeek Chat; and Google's Gemini 2.5 Flash and Nano Banana. Different models have different pricing, suitable for scenarios ranging from complex reasoning to large-scale batch processing.