Orpheus 3B
Orpheus 3B is an open-source text-to-speech model from Canopy Labs designed for expressive, natural-sounding speech synthesis.
Canopy Labs
DeepInfraExpressive open-source text-to-speech synthesis
Orpheus 3B is a text-to-speech model developed by Canopy Labs and released in March 2025 under the identifier canopylabs/orpheus-3b-0.1-ft. It is a 3-billion-parameter model hosted via DeepInfra and supports a context window of up to 10,000 tokens, making it suitable for converting longer passages of text into audio. The model is tagged as open source, meaning its weights are publicly available, and it is designed with expressiveness as a core characteristic of its output.
Orpheus 3B is best suited for applications that require natural, expressive speech generation from text input, such as voice assistants, audiobook narration, accessibility tools, and conversational agents. It accepts a voice selection input, allowing users to choose from available speaker voices. Because it is open source, developers can inspect, fine-tune, or self-host the model in addition to accessing it through providers like DeepInfra.
What Orpheus 3B supports
Text to Speech
Converts text input into spoken audio output, supporting up to 10,000 tokens of context per request.
Voice Selection
Allows users to choose from multiple available speaker voices via a select input at inference time.
Expressive Speech
Generates speech with natural prosody and expressiveness, a design focus reflected in the model's official tagging.
Open Source Weights
Model weights are publicly available, enabling fine-tuning, self-hosting, or inspection by developers.
Ready to build with Orpheus 3B?
Get Started FreeCommon questions about Orpheus 3B
What is the context window for Orpheus 3B?
Orpheus 3B supports a context window of 10,000 tokens, which determines the maximum length of text that can be processed in a single request.
Is Orpheus 3B free to use?
Pricing information is not published in the available metadata. Usage through DeepInfra may be subject to that provider's standard pricing. Check DeepInfra's pricing page for current rates.
Who developed Orpheus 3B?
Orpheus 3B was developed by Canopy Labs and released in March 2025. It is hosted and served via DeepInfra under the model identifier canopylabs/orpheus-3b-0.1-ft.
Can I fine-tune or self-host Orpheus 3B?
Yes. Orpheus 3B is open source, meaning its weights are publicly available. Developers can access them on Hugging Face to fine-tune or self-host the model.
What voice options are available in Orpheus 3B?
Orpheus 3B supports a voice selection input, allowing users to choose from available speaker voices at inference time. The specific list of voices depends on the deployment configuration via DeepInfra.
Documentation & links
Parameters & options
Voice to use in TTS.
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.
HexgradKokoro 82M
Kokoro 82M is an open-source text-to-speech model from Hexgrad supporting multiple languages with selectable 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.
OpenAIGPT-4o-mini TTS
GPT-4o-mini TTS is a text-to-speech model from OpenAI that converts text to audio with adjustable voice, speed, and format.
ElevenLabsElevenLabs TTS
ElevenLabs TTS is a text-to-speech model from ElevenLabs that converts text into natural-sounding audio speech.
Start building with Orpheus 3B
No API keys required. Create AI-powered workflows with Orpheus 3B in minutes — free.