> ## Documentation Index
> Fetch the complete documentation index at: https://docs.geko.sh/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI compatibility

> Point the OpenAI SDK at Geko by changing two lines.

Geko exposes an **OpenAI-compatible** speech endpoint at `POST /v1/audio/speech`. If you already use OpenAI's TTS, switch to Geko by changing the **base URL** and **model** — the rest of your code stays the same.

## Node (openai SDK)

```ts theme={null}
import OpenAI from "openai";
import { writeFile } from "node:fs/promises";

const client = new OpenAI({
  baseURL: "https://geko--tokay-serve-web.modal.run/v1",
  apiKey: process.env.GEKO_API_KEY, // your sk-tokay-... key
});

const res = await client.audio.speech.create({
  model: "tts-1",            // OpenAI ids are accepted + mapped to Geko's default
  voice: "Aigerim",          // a Geko voice name
  input: "Сәлеметсіз бе!",
  response_format: "wav",    // "wav" or "pcm"
});

await writeFile("hello.wav", Buffer.from(await res.arrayBuffer()));
```

## Python (openai SDK)

```python theme={null}
from openai import OpenAI

client = OpenAI(
    base_url="https://geko--tokay-serve-web.modal.run/v1",
    api_key=os.environ["GEKO_API_KEY"],
)

res = client.audio.speech.create(
    model="tts-1",
    voice="Aigerim",
    input="Сәлеметсіз бе!",
    response_format="wav",
)
res.stream_to_file("hello.wav")
```

## Notes & differences

* **Voices** are Geko voice names (`Aigerim`, `Arman`, …) — not OpenAI's (`alloy`, etc.). Fetch the live list from [`/v1/voices`](/voices).
* **`model`**: OpenAI ids (`tts-1`, `tts-1-hd`, `gpt-4o-mini-tts`) are accepted and mapped to Geko's default model; or pass a Geko model id like `tokay-kk-v1`.
* **`response_format`**: `wav` (default) and `pcm` (raw 16-bit PCM @ 24 kHz) are supported. `mp3`/`opus`/`aac`/`flac` return `400` for now — [convert with `ffmpeg`](/guides/playing-audio) or use the [native SDK](/sdk/typescript).
* For streaming and the full typed surface, prefer the [`@gekoai/sdk`](/sdk/typescript) client.

<Note>
  This endpoint is a thin compatibility shim over the same engine as [`POST /v1/tts`](/api/reference) — handy for dropping Geko into tools that already speak OpenAI.
</Note>

## Next steps

<CardGroup cols={2}>
  <Card title="TypeScript SDK" icon="code" href="/sdk/typescript">
    The native client with streaming and the full typed surface.
  </Card>

  <Card title="API reference" icon="server" href="/api/reference">
    The raw HTTP endpoints behind the compatibility shim.
  </Card>

  <Card title="Playing & saving audio" icon="volume-high" href="/guides/playing-audio">
    Turn the response bytes into files or playback.
  </Card>

  <Card title="Pick a voice" icon="waveform-lines" href="/voices">
    Geko voice names to pass as `voice`.
  </Card>
</CardGroup>