MiniMax Speech Turbo

Overview

MiniMax Speech Turbo converts text input into audio output with a focus on speed and efficiency. It supports multiple languages and operates in the cheap cost tier and fast speed tier, making it suitable for applications that prioritize low latency and high request volume over studio-grade fidelity. On Kyma, the model is accessible via an OpenAI-compatible endpoint using a single API key. It supports prompt caching, which bills repeated input prefixes at this model’s cached input rate, and includes automatic failover to reroute requests if a serving path degrades. Responses return exact billing metrics in the usage.cost field and report the active model in the X-Kyma-Model header. The endpoint does not support reasoning, vision, or structured outputs, and it is strictly a speech generation interface. It is designed for scalable deployments where consistent throughput and fast delivery are the primary requirements.

Specs

Field	Value
Model ID	`minimax-speech-turbo`
Best for	Real-time voice agents, conversational AI, bulk narration, high-throughput TTS
Context window	5,000
Input modalities	Text
Output modalities	Audio
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Fast
Cost band	Cheap
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.20
Output	$0.60

Use this when

Real-time voice agents — Generates rapid audio responses for interactive conversational systems.
Bulk narration pipelines — Processes large volumes of text into speech efficiently.
Conversational AI backends — Delivers fast, multilingual audio output for chat and voice assistants.
Scalable text to speech — Handles concurrent requests efficiently for high-volume audio generation.

Not ideal for

Do not use this model when you require studio-quality voice fidelity, complex audio editing, or structured JSON outputs.

Pick something else when

You need higher voice fidelity: use eleven-v3 or minimax-speech-hd.
You need music or sound effect generation: use elevenlabs-music or minimax-music-pro.
You need voice cloning or custom voice design: use minimax-voice-clone or minimax-voice-design.

Example

from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="minimax-speech-turbo",
    messages=[{"role": "user", "content": "..."}],
)

Agent query example

Ask the API which models fit, instead of hardcoding an id:

curl "https://kymaapi.com/v1/models?recommended_for=generate-speech&tools=true&quality_tier=efficient"

FAQ

Does this model support structured outputs or JSON formatting? No, it only accepts text input and returns audio output. How does prompt caching work for this endpoint? Repeated prompt prefixes are billed at this model’s cached input rate, reducing costs for recurring templates. What happens if the serving path experiences latency? Kyma automatically reroutes the request to a healthy serving path to maintain delivery.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ