MiniMax Speech HD

Overview

MiniMax Speech HD converts text into high-quality audio across multiple languages. It delivers expressive voice synthesis with a 5,000-token context window and is optimized for strong audio quality at a cheap tier. On Kyma, the model runs through an OpenAI-compatible endpoint using a single API key. Every request includes automatic failover if a serving path degrades, and responses return exact billing data in usage.cost alongside an X-Kyma-Model header. Prompt caching is supported, billing repeated prefixes at this model’s cached input rate. The model does not support reasoning, vision, or structured outputs. It operates at a medium speed tier, making it suitable for batch processing and asynchronous audio pipelines rather than low-latency conversational applications.

Specs

Field	Value
Model ID	`minimax-speech-hd`
Best for	Production narration, multilingual content, budget-tier brand voice
Context window	5,000
Input modalities	Text
Output modalities	Audio
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Medium
Cost band	Cheap
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.20
Output	$0.60

Use this when

Multilingual Content Narration — Generate expressive voiceovers for videos and podcasts across multiple languages.
Budget Brand Voiceovers — Produce consistent audio assets for marketing campaigns at a lower cost tier.
Audio Translation Workflows — Convert localized text into natural-sounding speech for global distribution.
Extended Audiobook Generation — Process long text passages into continuous audio using the full context window.

Not ideal for

It is not suitable for real-time conversational voice agents or applications requiring sub-second audio latency.

Pick something else when

You need faster audio generation for interactive voice agents: use eleven-turbo-v2-5 or minimax-speech-turbo.
You require custom voice cloning or specialized voice design: use minimax-voice-clone or minimax-voice-design.
You need premium-tier voice quality regardless of cost: use eleven-v3.

Example

from openai import OpenAI

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

response = client.chat.completions.create(
    model="minimax-speech-hd",
    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=strong"

FAQ

Does this model support prompt caching? Yes, Kyma caches repeated prompt prefixes and bills them at this model’s cached input rate. What happens if the serving path fails during generation? Kyma automatically reroutes the request to a healthy serving path without requiring client-side retries. Can I use this for real-time voice conversations? No, the model operates at a medium speed tier and is optimized for asynchronous or batch audio generation.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ