Qwen3 TTS Instruct Flash

Specs & limits

Instruction-controllable tier of Alibaba's Qwen3-TTS family on Model Studio, currently equivalent to the qwen3-tts-instruct-flash-2026-01-26 snapshot. Positioned for latency-tolerant work such as audiobook production, online education voiceovers and content creation. Priced in the Singapore region at $0.115 per 10,000 input characters in the international deployment scope, with a 110,000-character free quota valid for 90 days after activating Model Studio.

per Alibaba official docs ↗

Use Qwen3 TTS Instruct Flash in 30 seconds

OpenAI-compatible: swap the base_url, keep your SDK. POST /v1/chat/completions

from openai import OpenAI

client = OpenAI(
    base_url="https://synthorai.io/v1",
    api_key="sk-syn-...",
)

resp = client.chat.completions.create(
    model="qwen3-tts-instruct-flash",
    messages=[{"role": "user", "content": "Summarize this diff"}],
)
print(resp.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://synthorai.io/v1",
  apiKey: "sk-syn-...",
});

const resp = await client.chat.completions.create({
  model: "qwen3-tts-instruct-flash",
  messages: [{ role: "user", content: "Summarize this diff" }],
});
console.log(resp.choices[0].message.content);

curl https://synthorai.io/v1/chat/completions \
  -H "Authorization: Bearer sk-syn-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-tts-instruct-flash",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/option"
)

func main() {
	client := openai.NewClient(
		option.WithBaseURL("https://synthorai.io/v1"),
		option.WithAPIKey("sk-syn-..."),
	)
	resp, _ := client.Chat.Completions.New(context.TODO(), openai.ChatCompletionNewParams{
		Model: "qwen3-tts-instruct-flash",
		Messages: []openai.ChatCompletionMessageParamUnion{
			openai.UserMessage("Summarize this diff"),
		},
	})
	fmt.Println(resp.Choices[0].Message.Content)
}

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.chat.completions.*;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .baseUrl("https://synthorai.io/v1")
    .apiKey("sk-syn-...")
    .build();

ChatCompletion resp = client.chat().completions().create(
    ChatCompletionCreateParams.builder()
        .model("qwen3-tts-instruct-flash")
        .addUserMessage("Summarize this diff")
        .build());
System.out.println(resp.choices().get(0).message().content().orElse(""));

About Qwen3 TTS Instruct Flash

• Model Studio splits the family by transport rather than by capability: an id carrying the realtime suffix speaks WebSocket, while this one, without it, is the HTTP model behind what Alibaba calls non-real-time speech synthesis, designed for latency-tolerant scenarios such as audiobook production, online education voiceovers, and content creation.
• Instruction control is the reason to reach for it over plain qwen3-tts-flash.
• You describe the delivery you want in ordinary language to control speed, emotion, and style per request, the documented examples being to speak gently at a slower pace or to use an excited broadcast style; the instructions field takes up to 1,600 tokens and is documented for Chinese and English only, and optimize_instructions, off by default, has the service rewrite your wording into a directive better suited to synthesis.
• Voices are the family's named personas, among them Cherry as a sunny, positive, friendly, and natural young woman and Nofish as a designer who cannot pronounce retroflex sounds, and the voice list states which model ids accept each one.
• Ten languages are covered, Chinese (Mandarin), English, German, Italian, Portuguese, Spanish, Japanese, Korean, French, and Russian, and unlike qwen3-tts-flash the Instruct line carries none of the Chinese dialect voices.
• Two limits shape integration: input text is capped at 600 characters, and language_type defaults to Auto, which Alibaba says does not guarantee accuracy and advises replacing with an explicit language for single-language text.
• Non-streaming returns an audio file URL valid for 24 hours, while streaming returns Base64-encoded PCM in chunks with the URL only in the final packet, and the official samples play that stream back as 24 kHz mono 16-bit audio.
• Billing counts input text characters, at $0.115 per 10,000 in the international deployment scope, and output audio is free.
• Synthorai serves it through the OpenAI-compatible /v1/audio/speech endpoint.

FAQ

Related models

Compare