nimbus.tools
Get an API key

Continue

Continue.dev is the open-source pair-programmer for VS Code and JetBrains. Its config.json takes any OpenAI-compatible provider, so Nimbus registers with a single entry.

Continue treats models as a list where each entry declares a provider, an API base, and an API key. Pointing the provider at openai plus setting apiBase to Nimbus makes every Nimbus model available in the chat side-panel, inline edits, and autocomplete.

Prerequisites

  • Continue extension installed in VS Code or JetBrains.
  • A Nimbus API key (sk-nim-...).
  • ~/.continue/config.json reachable — open the file with Continue: Open config.json from the command palette.

Install

Install from the marketplace:

  • VS Code: Continue.continue
  • JetBrains: search "Continue" in the plugin marketplace

Configure

Add entries under models in ~/.continue/config.json. Set provider to openai and apiBase to the Nimbus /v1 endpoint.

json
{
  "models": [
    {
      "title": "Nimbus · Opus 4.8",
      "provider": "openai",
      "model": "anthropic/claude-opus-4.8",
      "apiBase": "https://llm.nimbusapi.net/v1",
      "apiKey": "sk-nim-..."
    },
    {
      "title": "Nimbus · Sonnet 4.6",
      "provider": "openai",
      "model": "anthropic/claude-sonnet-4.6",
      "apiBase": "https://llm.nimbusapi.net/v1",
      "apiKey": "sk-nim-..."
    },
    {
      "title": "Nimbus · GPT-5.1",
      "provider": "openai",
      "model": "openai/gpt-5.1",
      "apiBase": "https://llm.nimbusapi.net/v1",
      "apiKey": "sk-nim-..."
    }
  ],
  "tabAutocompleteModel": {
    "title": "Nimbus · Autocomplete",
    "provider": "openai",
    "model": "openai/gpt-5-mini",
    "apiBase": "https://llm.nimbusapi.net/v1",
    "apiKey": "sk-nim-..."
  }
}

Save the file — Continue hot-reloads. The picker in the chat side-panel now shows each title.

First call

Open the Continue side-panel (Cmd/Ctrl-L), pick Nimbus · Opus 4.8 from the model dropdown, and send a prompt. The request goes to llm.nimbusapi.net/v1/chat/completions and streams back. Confirm delivery in the Nimbus dashboard's Requests view.

For inline edits, highlight code, press Cmd/Ctrl-I, and pick the same model — the diff view uses the streamed response directly.

Troubleshooting

  • Model absent from picker after save: Continue rejected the config as invalid — check the status bar for a red dot and open Continue: View Logs for the parse error. Usually a trailing comma or a smart quote.
  • 401 Unauthorized: apiKey is still the literal sk-nim-... placeholder, or the key was rotated. Copy a fresh one from the Nimbus dashboard.
  • 404 Not Found: apiBase is missing the /v1 suffix. Continue passes apiBase through verbatim — no implicit suffix.
  • Autocomplete flickers off: tabAutocompleteModel is pointing at a slow reasoning model. Route it at a fast small model like openai/gpt-5-mini or anthropic/claude-haiku-4.
  • "Model not found" in the logs: you used a bare name like gpt-5.1 in the model field. Nimbus requires the provider prefix. Full catalog at Models.