Continue
Open-source autopilot — set an OpenAI-compatible provider in config.yaml.
VS Code's AI extensions speak the OpenAI API. Point any of them at ubc now's endpoint and get frontier models in your editor on a free daily budget.
Any VS Code extension that accepts a custom OpenAI base URL works with ubc now. The popular ones:
Open-source autopilot — set an OpenAI-compatible provider in config.yaml.
Agentic coding — choose the “OpenAI Compatible” provider (see the Cline guide).
Cline forks that take the same OpenAI-compatible base URL + key.
Mint an API key at /api-keys after verifying with Self. Keep it handy; it starts with ubc_sk_.
Continue is the simplest starting point. Install it from the VS Code Marketplace, then open its config.
In Continue's config.yaml, add a model with the openai provider and ubc now's apiBase. Replace the model id with one from /v1/models.
models:- name: ubc nowprovider: openaiapiBase: https://ubcnow.ai/v1apiKey: ubc_sk_…model: <model-id>
Reload VS Code, open the extension's chat, select the ubc now model, and send a prompt.
Continue's `model` field must be a real ubc now model id. List them first.
curl https://ubcnow.ai/v1/models \-H "Authorization: Bearer ubc_sk_…"
ubc now keys (ubc_sk_…) work anywhere an OpenAI key is accepted — the prefix differs but the Bearer-token auth is identical.
Your key must be an active ubc now key (it starts with ubc_sk_). Mint or re-mint one at /api-keys, and make sure it is sent as a Bearer token, not pasted into a username field.
You have spent your daily token allowance. It resets at 00:00 UTC. Image and chat usage both draw from the same daily pool.
Use an exact model id from https://ubcnow.ai/v1/models — copy it verbatim. ubc now does not alias OpenAI model names like "gpt-4o"; pick one of the models it actually serves.
Mint a key and point your favorite VS Code extension at ubc now.