This project provides a reverse-engineered, free API for the YueWen Step (StepChat) multimodal LLM, offering high-speed streaming, web search, document/image analysis, and multi-turn conversations. It targets developers and researchers needing to integrate StepChat's advanced multimodal capabilities via a familiar, ChatGPT-compatible interface, bypassing direct official SDK integration.

How It Works

The project acts as a proxy, replicating API calls from the official StepChat web interface. It requires user-obtained deviceId and Oasis-Token credentials from yuewen.cn for authentication. Its primary advantage is the OpenAI-compatible /v1/chat/completions endpoint, enabling seamless integration with existing LLM tooling. It supports multimodal inputs (documents, images) and streaming outputs, mirroring official platform functionality.

Quick Start & Requirements

• Installation: Deployment options include Docker, Docker-compose, Render, Vercel, or native Node.js.
• Prerequisites: Requires deviceId and Oasis-Token from yuewen.cn (obtained via browser dev tools). Node.js needed for native builds. Multi-account access via comma-separated tokens.
• Resource Footprint: Docker requires a public IP and port 8000. Free tiers on Render/Vercel may have limitations (e.g., timeouts, idle sleep).
• Links: Official StepChat: https://platform.stepfun.com/. Clients: LobeChat (https://github.com/Yanyutin753/lobe-chat), ChatGPT Web (https://github.com/SuYxh/chatgpt-web-sea).

Highlighted Details

• Multimodal Capabilities: Advanced image and long document analysis alongside text chat.
• OpenAI Compatibility: /v1/chat/completions endpoint mirrors OpenAI's API.
• Deployment Flexibility: Supports Docker, serverless platforms, and native Node.js.
• Streaming Support: Enables high-speed, real-time streaming output.

Maintenance & Community

Maintained by LLM-Red-Team, this project is part of a series of reverse-API wrappers for various LLMs, suggesting an active community focus. No specific community links or roadmap are detailed.

Licensing & Compatibility

• License: Not explicitly stated. Designated for "testing only" and "personal use," with explicit prohibitions against external service provision or commercial use.
• Compatibility: Compatible with OpenAI API clients. Commercial use is discouraged due to its reverse-engineered nature and potential instability.

Limitations & Caveats

• Instability: Labeled an "unstable reverse API"; users advised to use the official paid service to avoid risks like API bans.
• Terms of Service: Restricted to personal testing/learning; external service provision or commercial use is prohibited and carries risks.
• Token Counting: Token usage is not accurately tracked; fixed numbers are returned.
• Dependency: Relies on the continued availability and structure of the yuewen.cn web interface and its authentication tokens.