Kokoro 82M
Kokoro 82M is an open-source text-to-speech model from Hexgrad supporting multiple languages with selectable voices.
Hexgrad
DeepInfraOpen-source multilingual text-to-speech synthesis
Kokoro 82M is a text-to-speech model developed by Hexgrad and released in January 2025. It has 82 million parameters and is designed to convert text into natural-sounding speech, with support for multiple languages and a context window of 10,000 tokens. The model is open-source and available through providers including DeepInfra.
Kokoro 82M is suited for applications that require multilingual voice synthesis without the overhead of much larger models. It offers selectable voice options, making it adaptable for use cases such as content narration, accessibility tooling, and voice interface prototyping. Its open-source nature and low-cost availability make it accessible for developers building speech-enabled applications.
What Kokoro 82M supports
Text to Speech
Converts input text into synthesized audio speech. Supports a context window of up to 10,000 tokens per request.
Multilingual Support
Generates speech across multiple languages from a single model. Enables voice synthesis for international or multilingual applications.
Voice Selection
Allows users to choose from multiple available voice options via a select input. Provides control over speaker identity for different use cases.
Open-Source Model
The model weights are publicly available under an open-source license on Hugging Face. Developers can inspect, download, or self-host the model.
Low-Cost Inference
Tagged as a low-cost model, making it suitable for high-volume or budget-conscious speech synthesis workloads.
Ready to build with Kokoro 82M?
Get Started FreeCommon questions about Kokoro 82M
What is the context window for Kokoro 82M?
Kokoro 82M supports a context window of 10,000 tokens per request.
How much does it cost to use Kokoro 82M?
Kokoro 82M is tagged as a low-cost model. Specific per-character or per-request pricing is determined by the provider, DeepInfra. Check DeepInfra's pricing page for current rates.
Is Kokoro 82M open source?
Yes, Kokoro 82M is open source. The model weights are publicly available on Hugging Face under the hexgrad/Kokoro-82M repository.
What languages does Kokoro 82M support?
Kokoro 82M is tagged as multilingual, meaning it supports speech synthesis in more than one language. Refer to the Hugging Face model card for the specific list of supported languages.
Can I choose different voices with Kokoro 82M?
Yes, the model exposes a voice selection input, allowing you to pick from available speaker voices when making a request.
When was Kokoro 82M released?
Kokoro 82M was released in January 2025 by Hexgrad.
Parameters & options
Voice to use in TTS. The prefix encodes language and gender.
Explore similar models
QwenQwen3 TTS
Qwen3 TTS is an open-source multilingual text-to-speech model from Qwen with a 10,000-token context window.
Canopy LabsOrpheus 3B
Orpheus 3B is an open-source text-to-speech model from Canopy Labs designed for expressive, natural-sounding speech synthesis.
OpenAITTS-1
TTS-1 is a text-to-speech model from OpenAI that converts written text into natural-sounding audio across multiple voices.
GoogleGemini 3.1 Flash TTS
Gemini 3.1 Flash TTS is a text-to-speech model from Google that converts text into audio with style control.
MiniMaxMinimax Speech 2.8 HD
Minimax Speech 2.8 HD is a text-to-speech model from MiniMax supporting emotion, pitch, speed, and multi-format audio output.
CartesiaCartesia Sonic 3
Start building with Kokoro 82M
No API keys required. Create AI-powered workflows with Kokoro 82M in minutes — free.