Switch active TTS model
Save the preferred TTS model family and return its capabilities.
Unloads the current model. With auto_load set, the new model starts
loading right away.
application/json- body
model_family*Model FamilyModel family to switch to
auto_load?Auto LoadIf true, immediately start downloading/loading the model
falsemode?|Generation mode ('clone'/'design'/'tts') to persist this family as the preference for
Successful Response
application/json- response
model_family*Model FamilyModel family identifier
display_name*Display NameHuman-readable model name
supports_clone*Supports CloneWhether the model supports voice cloning
supports_design*Supports DesignWhether the model supports voice design
supports_custom_voice*Supports Custom VoiceWhether the model supports custom voice speakers
supports_streaming*Supports StreamingWhether the model supports streaming generation
supports_emotional_clone?Supports Emotional CloneWhether the model can clone with an emotion and intensity
falsesupports_design_reference?Supports Design ReferenceWhether voice design can be conditioned on a saved reference voice
falsedesign_takes_target_text?Design Takes Target TextFalse when design mode has no text to speak (the model writes the words)
truesample_rate*Sample RateOutput audio sample rate in Hz
ref_min_sec?Ref Min SecShortest reference audio the model accepts, in seconds
1.5ref_min_speech_sec?Ref Min Speech SecLeast speech the reference must contain, in seconds
3ref_max_sec?Ref Max SecLongest reference audio the model uses, in seconds
10custom_voice_speakers*array<string>Available predefined speakers
generation_params*array<>Available generation parameters with ranges
variants*array<>Available model variants/quality tiers
studio_only?Studio OnlyTrue if the model is only offered in the Studio audiobook tools
falseloading?|True when switching started a background model load
curl -X PUT 'http://127.0.0.1:8100/api/v1/models/active' \ -H 'Content-Type: application/json' \ -d '{ "model_family": "qwen3-tts"}'{ "model_family": "string", "display_name": "string", "supports_clone": true, "supports_design": true, "supports_custom_voice": true, "supports_streaming": true, "supports_emotional_clone": false, "supports_design_reference": false, "design_takes_target_text": true, "sample_rate": 0, "ref_min_sec": 1.5, "ref_min_speech_sec": 3, "ref_max_sec": 10, "custom_voice_speakers": [ "string" ], "generation_params": [ {} ], "variants": [ {} ], "studio_only": false, "loading": true}