SDK / Integrations/App integrations

Claude Code

Two environment variables point Claude Code at APIONE — and any vendor's model.

On this page

Claude Code speaks the Anthropic protocol, so ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN are all it takes.

Configuration

bash
export ANTHROPIC_BASE_URL="https://api.apione.ai"
export ANTHROPIC_AUTH_TOKEN="YOUR_APIONE_KEY"

claude

Put them in ~/.zshrc or ~/.bashrc to avoid re-exporting. With direnv, a per-project .envrc is cleaner still.

Choosing models

bash
export ANTHROPIC_MODEL="claude-opus-4-7"
export ANTHROPIC_SMALL_FAST_MODEL="claude-haiku-4-5-20251001"

Use plain model IDs. You can also point at another vendor entirely (glm-5.3) — the gateway translates.

Verify

bash
curl -s https://api.apione.ai/v1/messages \
  -H "x-api-key: $ANTHROPIC_AUTH_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-haiku-4-5-20251001","max_tokens":16,
       "messages":[{"role":"user","content":"ping"}]}'

If content comes back, the configuration is right — start Claude Code.

Troubleshooting

SymptomFix
Still hitting the official APIVariables not in scope — run echo $ANTHROPIC_BASE_URL in the same shell
404An extra /v1 in the base URL
400 model_not_supportedModel ID misspelled or not available
401Set ANTHROPIC_API_KEY instead of ANTHROPIC_AUTH_TOKEN, or both are set and conflict
Note

Once the key points at APIONE, usage lands on your APIONE bill — usage can show this machine's spend on its own key.

Contact us

Something unclear, or need a human? Pick the channel that fits your question.