Claude Code

Claude Code is Anthropic’s agentic coding tool that runs in the terminal. By default it uses Anthropic models, but it supports any OpenAI-compatible provider through a custom API endpoint. You can route Claude Code through LiteLLM — a lightweight proxy that translates OpenAI-format requests — to use CompactifAI models for coding assistance.

You need a CompactifAI API key and the model id for each model you plan to use (for example hypernova-60b or glm-5-1). Use CompactifAI API https://api.compactif.ai/v1 unless you are explicitly targeting another environment.

How it works

Claude Code  -->  LiteLLM Proxy  -->  CompactifAI API

Claude Code sends requests in OpenAI chat-completion format to a local LiteLLM proxy. LiteLLM forwards them to CompactifAI’s /v1/chat/completions endpoint and streams the response back.

1. Install Claude Code

If you don’t have Claude Code installed yet, run:

curl -fsSL https://claude.ai/install.sh | bash
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc
claude --version

2. Install LiteLLM

pip install 'litellm[proxy]'

3. Create a LiteLLM config file

Create a litellm_config.yaml in your project (or anywhere on disk):

model_list:
  - model_name: hypernova-60b
    litellm_params:
      model: compactifai/hypernova-60b
      api_base: https://api.compactif.ai/v1
      api_key: os.environ/COMPACTIFAI_API_KEY
  - model_name: glm-5-1
    litellm_params:
      model: compactifai/glm-5-1
      api_base: https://api.compactif.ai/v1
      api_key: os.environ/COMPACTIFAI_API_KEY
  - model_name: llama-4-scout
    litellm_params:
      model: compactifai/llama-4-scout
      api_base: https://api.compactif.ai/v1
      api_key: os.environ/COMPACTIFAI_API_KEY
litellm_settings:
  drop_params: true

Replace the model_name entries with any models from the models catalog you have access to. You can discover the full list of available model IDs programmatically via GET /v1/models (see API reference). The litellm_params.model value must be prefixed with openai/ so LiteLLM uses the OpenAI-compatible path.

4. Set your API key

Export your CompactifAI API key as an environment variable (the config above references it via os.environ/):

export COMPACTIFAI_API_KEY="your-compactifai-api-key"

5. Start the LiteLLM proxy

litellm --config litellm_config.yaml --port 4000

You should see output confirming the proxy is running on http://localhost:4000.

6. Configure Claude Code

Claude Code reads its provider settings from environment variables. Set the following before launching Claude Code:

export CLAUDE_CODE_USE_BEDROCK=0
export ANTHROPIC_BASE_URL="http://localhost:4000"
export ANTHROPIC_API_KEY="dummy"

Then launch Claude Code:

claude

7. Select a CompactifAI model

Inside Claude Code, use the /model command to switch to one of the model names you defined in litellm_config.yaml:

/model hypernova-60b

Or set it via environment variable before starting:

export ANTHROPIC_MODEL="hypernova-60b"
claude

Claude Code will now send all coding requests through LiteLLM to CompactifAI.

8. Verify it works

Ask Claude Code a simple question — for example, type “list the files in this directory”. If the response streams back, the integration is working. Requests use the same /v1/chat/completions flow described in Chat Completion.

Troubleshooting

Issue	Solution
`Connection refused` on localhost:4000	Make sure the LiteLLM proxy is running (`litellm --config litellm_config.yaml --port 4000`)
`401 Unauthorized` from CompactifAI	Verify `COMPACTIFAI_API_KEY` is set and matches the key from your MultiverseIAM dashboard
Model not found error	Ensure the `model_name` in `litellm_config.yaml` matches what you pass to `/model` in Claude Code, and that the underlying model id (e.g. `hypernova-60b`) exists in the models catalog
Claude Code ignores the base URL	Confirm `ANTHROPIC_BASE_URL` is exported before launching `claude`. Also ensure `CLAUDE_CODE_USE_BEDROCK` is `0`

For request and response fields, see Chat Completion and the API reference.