Continue's chat, edit and apply, on your VelATL balance.

Setup - config.yaml

Create a key in Settings → API keys first (new keys start capped at 5,000 credits/month), then add both entries:

models:
  - name: VelATL Swift
    provider: openai
    model: fast
    apiBase: https://velatl.ai/api/v1
    apiKey: velatl_...
    roles: [chat, edit, apply]
  - name: VelATL Balanced
    provider: openai
    model: balanced
    apiBase: https://velatl.ai/api/v1
    apiKey: velatl_...
    roles: [chat, edit, apply]

Keep the fast entry first: use Swift for chat and quick edits, switch to Balanced per task for harder refactors. Every role bills at the same per-token rates - see them live via GET /api/v1/models.

What works - and what is rejected

• Streaming, autocomplete-style edits and apply flows work; text + tools only.
• Image inputs, forced tool choice and response_format combined with tools are rejected with a clear 400 before billing.
• Out of credits → 402 insufficient_credits - top up in billing; failed calls refund automatically.

Other setups: OpenCode, Cline, Claude Code. Full protocol reference: developer docs.