Introduction to AI-VERDE API¶
AI-VERDE exposes an OpenAI-compatible API, so any tool or library that supports OpenAI can connect to it using your AI-VERDE API key. AI-VERDE also accepts Anthropic-style requests (used by Claude Code and the Anthropic SDK) and, for Claude Code only, AWS Bedrock-style requests for Claude models hosted on AWS Bedrock. This section explains how to get your key and use it across a range of environments.
Getting Started¶
Before using the API you will need to:
- Obtain your API key — Sign in at chat.cyverse.ai, open your course or team details, and copy the key from the API Key tab. See Obtaining your AI-VERDE API Key for step-by-step instructions.
- Know which models are available — Your available models are listed on the same API Key page. You can also retrieve them programmatically; see Getting a List of Models.
What's in This Section¶
| Section | What it covers |
|---|---|
| Using Your API Key for AWS Bedrock Models | Claude models on AWS Bedrock with Claude Code and the Claude Code for VS Code extension |
| Using Your API Key in Code | Python integration with LangChain, LlamaIndex, and the Anthropic Python SDK |
| Using CLI Coding Assistants | Terminal-based AI coding tools: Claude Code Router, Aider, Claude Code, Codex, and OpenCode |
| Using IDEs | VS Code integration: Claude Code for VS Code and Codex for VS Code |
| Using Desktop Clients | GUI chat clients such as Chatbox AI |
Base URL and Authentication¶
When configuring a client, you need your API key (copied from the AI-VERDE dashboard) and a base URL. The base URL depends on which style of API the client speaks:
| API style | Base URL | Used by |
|---|---|---|
| OpenAI-compatible | https://llm-api.cyverse.ai/v1 |
Aider, Codex, OpenCode, Claude Code Router, Chatbox AI, Codex for VS Code, and most OpenAI clients |
| Anthropic-style | https://llm-api.cyverse.ai |
Claude Code (Anthropic models), Claude Code (non-Anthropic models), Claude Code for VS Code, Anthropic Python SDK |
| AWS Bedrock | https://llm-api.cyverse.ai/bedrock |
Claude Code on Bedrock, Claude Code for VS Code on Bedrock |
If your course details show a different base URL ending in /v1, use it for OpenAI-compatible clients, remove /v1 for Anthropic-style clients, and replace /v1 with /bedrock for AWS Bedrock.
Note
The LangChain and LlamaIndex examples use LiteLLM's litellm_proxy provider, which takes the base URL without /v1 (https://llm-api.cyverse.ai).
OpenAI-compatible clients send the API key as a bearer token (Authorization: Bearer <your-api-key>), and Anthropic-style clients may send it in an x-api-key header instead; AI-VERDE accepts either. Each guide shows which setting or environment variable holds the key for that tool.