Phonon 31M
Phonon 31M is available now for general LLM work with open-source; evaluate it while provider pricing coverage matures.
- Family
- Phonon
- Released
- 2026-10-20
- Parameters
- 31M
- Specialization
- text-to-speech
- Openness
- Open source
- License
- Apache 2.0OSI-approvedPermitted
- Weights
- Available
- Code
- Unknown
No tracked provider pricing is available yet.
About
Phonon 31M is Gradium's smallest open on-device text-to-speech model, released October 20, 2026 under Apache 2.0. It has 31 million parameters (excluding the audio tokenizer), speaks English as streaming 24 kHz mono audio, and runs on laptop CPUs, phones, browsers and GPUs; Gradium says it starts speaking in about 7 ms on a laptop CPU. It builds on Kyutai's Pocket TTS and Mimi codec, and is distilled from a larger Phonon teacher. Weights (safetensors and 8-bit/4-bit GGUF, about 64 MB) are on Hugging Face.
Providers
No active provider routes are tracked for this model yet.
Capabilities
Benchmark peer barsfor Coding
No task-mapped benchmark peers are available for this model yet.
Migration checks
No linked migration route is available for this model yet.
API versions
gradium/phonon-31m-enphonon-31m-enNo tracked provider pricing is available yet.