Custom Model Option - Claude Code
Add gateway or custom models to the picker via environment variables.
Custom model options let you point Claude Code at an alternative endpoint - an LLM gateway, a Bedrock or Vertex deployment, or a proxy - without leaving the CLI.
What it does
You set environment variables to describe a custom model (name, endpoint, headers, model ID). That option shows up in the /model picker alongside the defaults. Claude Code sends the requests through your configured path but keeps the rest of the session behavior identical. It's how teams with internal gateways keep Claude Code on corporate rails.
When to use it
- Routing through LiteLLM, LangFuse, or a custom gateway for observability.
- Team deployments that require Bedrock, Vertex, or Azure Foundry access.
- Testing a model version that isn't in the default picker.
- Swapping between clouds without editing multiple config files.
Gotchas
- Gateway auth is your responsibility. Wrong headers mean silent 401s.
- Some features (prompt caching, channels) depend on Anthropic-direct endpoints.
- Model aliases don't transparently resolve on custom endpoints. Use full IDs.
Official docs: https://code.claude.com/docs/en/model-config.md#add-a-custom-model-option
Technical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
Get the weekly deep dive
Tutorials on Claude Code, AI agents, and dev tools - delivered free every week.
Was this helpful?
