@qubittron/bastion-sdk

Official TypeScript SDK for the Bastion sovereign Canadian LLM API.

Alpha — v0.1.3. API surface may change before 1.0.

Install

npm install @qubittron/bastion-sdk
# or
bun add @qubittron/bastion-sdk

Quickstart

import { Bastion } from "@qubittron/bastion-sdk";

// reads BASTION_API_KEY from env if `apiKey` is omitted
const client = new Bastion({ apiKey: process.env.BASTION_API_KEY });

const res = await client.chat.completions.create({
  model: "gpt-oss-120b",
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(res.choices[0]?.message.content);

Streaming

const stream = await client.chat.completions.create({
  model: "gpt-oss-120b",
  messages: [{ role: "user", content: "Tell me a poem" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta.content ?? "");
}

List models

const { data } = await client.models.list();
for (const m of data) console.log(m.id);

Images

const res = await client.images.generate({
  model: "sd-xl",
  prompt: "a sovereign Canadian moose, oil painting",
  size: "1024x1024",
  n: 1,
});
console.log(res.data[0]?.url);

Speech (TTS)

Synthesize speech via the NVIDIA Riva proxy. Returns the raw audio bytes plus the upstream Content-Type so you know which extension to write.

import { writeFile } from "node:fs/promises";
import { SpeechEncoding } from "@qubittron/bastion-sdk";

const { audio, contentType } = await client.audio.speech({
  text: "Bonjour le monde",
  language_code: "fr-CA",
  voice_name: "French-Canadian.Female-1",
  encoding: SpeechEncoding.LINEAR_PCM, // 1
  sample_rate_hz: 44100,
});
console.log(contentType); // e.g. "audio/wav"
await writeFile("hello.wav", audio);

Transcription (STT)

Transcribe audio. Accepts a Blob or File and posts as multipart/form-data.

import { readFile } from "node:fs/promises";

const buf = await readFile("speech.wav");
const file = new File([buf], "speech.wav", { type: "audio/wav" });

const res = await client.audio.transcriptions.create({
  file,
  model: "whisper-large-v3",
  response_format: "verbose_json",
});
console.log(res.text);

Configuration

new Bastion({
  apiKey: "sk-...",                              // required
  baseURL: "https://bastion.qubittron.ai",       // optional (default)
  fetch: customFetch,                            // optional (defaults to global fetch)
  defaultHeaders: { "x-trace-id": "..." },       // optional
});

If apiKey is omitted, the SDK reads BASTION_API_KEY from process.env.

Error handling

All API errors extend BastionError and expose status, code, type, and the raw body.

import {
  AuthenticationError,
  BadRequestError,
  PermissionDeniedError,
  RateLimitError,
  UpstreamError,
  APIConnectionError,
} from "@qubittron/bastion-sdk";

try {
  await client.chat.completions.create({ model: "x", messages: [] });
} catch (err) {
  if (err instanceof RateLimitError) {
    // back off
  } else if (err instanceof AuthenticationError) {
    // refresh key
  } else if (err instanceof APIConnectionError) {
    // network problem
  } else {
    throw err;
  }
}

| HTTP | Class | |---|---| | 400 | BadRequestError | | 401 | AuthenticationError | | 402, 403 | PermissionDeniedError | | 404 | NotFoundError | | 429 | RateLimitError | | 502, 503, 504 | UpstreamError | | other | APIError | | network | APIConnectionError |

Compatibility

• Node 20+ (uses global fetch and ReadableStream)
• Bun
• Edge runtimes (Cloudflare Workers, Vercel Edge)
• Modern browsers with global fetch

Module formats: ESM + CJS. Types included.

Scope (v0.1)

• ✅ chat.completions.create (sync + stream)
• ✅ models.list
• ✅ images.generate
• ✅ audio.speech, audio.transcriptions.create
• ⬜ embeddings, responses (coming)
• ⬜ automatic retries / backoff (coming)
• ⬜ tool / function calling helpers (passthrough works today)

License

MIT