Use Eleven Multilingual V2 with Comfy Router

Call elevenlabs/eleven_multilingual_v2 through Comfy Router: endpoint, request shape and the response Router returns.

API Reference for elevenlabs/eleven_multilingual_v2, served by Comfy Router from Elevenlabs.

Input: Text · Output: Audio

Quick start

Create a key in your Comfy workspace and export it as COMFY_API_KEY. The Python and TypeScript snippets use the Comfy SDKs (pip install comfy-sdk and npm install @comfyorg/sdk); the cURL snippet is the same call over raw HTTP.

Model ID: elevenlabs/eleven_multilingual_v2

Endpoint: POST https://api.comfy.org/v2/models/elevenlabs/eleven_multilingual_v2

Wait for the result

Python

from comfy_sdk import Comfy

# Reads COMFY_API_KEY from the environment.
# The SDK automatically creates an idempotency key and reuses it for automatic retries.
with Comfy() as client:
    result = client.models.run(
        "elevenlabs/eleven_multilingual_v2",
        {
            "text": "Hello from Comfy Router.",
            "voice_id": "21m00Tcm4TlvDq8ikWAM",
        },
    )

print(result)

Queue and collect later

The same body, sent to POST https://api.comfy.org/v2/models/elevenlabs/eleven_multilingual_v2/requests. Router answers 201 with a request_id as soon as the run is admitted, and the result is collected once it is ready, from this process or another one. Queued delivery walks through status, cancellation and collection.

Python

import asyncio
from comfy_sdk import AsyncComfy

# Reads COMFY_API_KEY from the environment.
# Each submit() call mints its own Idempotency-Key and reuses it for automatic retries.
async def main():
    async with AsyncComfy() as client:
        handle = await client.models.submit(
            "elevenlabs/eleven_multilingual_v2",
            {
                "text": "Hello from Comfy Router.",
                "voice_id": "21m00Tcm4TlvDq8ikWAM",
            },
        )
        print("request_id:", handle.request_id)  # with the model ID, all another process needs

        # Poll until the request completes, waiting the Retry-After the server names.
        async for update in handle.iter_events():
            print(update.status, update.queue_position)

        # The provider's own payload, the same value models.run() returns.
        # A request that failed or was cancelled raises the typed Router error here.
        result = await handle.get()

    print(result)

asyncio.run(main())

Schema

Input

apply_language_text_normalization · boolean · default: "false"

Controls language-specific text normalization. Can heavily increase latency. Currently only supported for Japanese.

apply_text_normalization · string · default: "\"auto\""

Controls text normalization. ‘auto’ lets the system decide, ‘on’ always applies normalization, ‘off’ skips normalization.

Possible values: auto, on, off

language_code · string

Language code (ISO 639-1) to enforce for the model. If unsupported, an error is returned.

model_id · string · default: "\"eleven_multilingual_v2\""

Identifier of the model to use. Query /v1/models to list available models.

next_request_ids · string[]

Request IDs of next generations for continuity. Maximum 3.

next_text · string

Text that comes after this request, used to improve speech continuity.

previous_request_ids · string[]

Request IDs of previous generations for continuity. Maximum 3.

previous_text · string

Text that came before this request, used to improve speech continuity.

pronunciation_dictionary_locators · object[]

List of pronunciation dictionary locators (id, version_id). Maximum 3 per request.

pronunciation_dictionary_locators[].pronunciation_dictionary_id · string · required

The ID of the pronunciation dictionary

pronunciation_dictionary_locators[].version_id · string · required

The version ID of the pronunciation dictionary

seed · integer

Seed for deterministic generation. Must be between 0 and 4294967295.

Range: 0 to 4294967295

text · string · required

The text that will be converted into speech.

use_pvc_as_ivc · boolean · default: "false"

Deprecated. If true, uses IVC version of voice instead of PVC.

voice_settings · object

Voice settings configuration

voice_settings.similarity_boost · number · default: "0.75"

How closely the AI adheres to the original voice when replicating it.

Range: 0 to 1

Format: double

voice_settings.speed · number · default: "1"

Speed adjustment. 1.0 is default, values below slow down, values above speed up.

Range: 0.7 to 1.2

Format: double

voice_settings.stability · number · default: "0.5"

Stability of the voice. Lower values introduce broader emotional range.

Range: 0 to 1

Format: double

voice_settings.style · number · default: "0"

Style exaggeration. Amplifies the style of the original speaker.

Range: 0 to 1

Format: double

voice_settings.use_speaker_boost · boolean · default: "true"

Boosts similarity to the original speaker. Requires higher computational load.

voice_id · string · required

ID of the voice to use, letters and digits only. Router places it in the ElevenLabs request path.

Generated from the schema Router serves at GET /v2/models/elevenlabs/eleven_multilingual_v2/openapi.json, the same document it validates a call against before the request reaches the provider.

Output

*/* · string (binary)

Raw audio bytes. The Content-Type and encoding follow the requested output_format and are forwarded from ElevenLabs. The example is a placeholder for the binary body, not JSON or base64.

Examples

Input

{
  "text": "Hello from Comfy Router.",
  "voice_id": "21m00Tcm4TlvDq8ikWAM"
}

Output

Binary body: raw bytes rather than a JSON document, returned as */*, so there is no JSON example to show. The response Content-Type and encoding follow the request, as the Output schema above describes.

Before you ship

The SDKs create an Idempotency-Key and reuse it for automatic retries. For manual retries, reuse the original key. Router can hold the connection for up to 10 minutes.

When a request fails, Router sends an X-Comfy-Error-Type response header explaining why. A 422 means Router rejected the input before calling the provider, and a 413 means the request body was larger than Router accepts. Download generated assets promptly because result URLs can expire.

Any size limit named in a field description above is the provider’s own bound on that field, quoted from the provider’s specification. Router applies a separate cap to the whole request body, which base64-encoded media counts against: see request body size.

This page documents one partner model called through Comfy Router. The same comfy-sdk / @comfyorg/sdk package also ships a second client, for running a whole ComfyUI workflow graph on Comfy Cloud: Comfy(api_key=...) / new Comfy({ apiKey }), with client.workflows, client.assets and client.jobs. See Comfy SDKs.

Headers

Authentication, idempotency, request IDs, error buckets, retry pacing, spend limits.

Using the Router API

Model discovery, validation errors, retries, and billing.

Limitations

What Router does not do today, and what to use instead.