Liquid AI
S2S 3 models no stock voices based in United States self-serve API data from documentation synced 2026-11-18
Foundation-model company spun out of MIT CSAIL, based in Cambridge, Massachusetts (US). Builds LFM (Liquid Foundation Models) for on-device and hosted use, including end-to-end audio models (LFM3-Audio, LFM3-Omni) that do ASR, TTS, speech translation and spoken chat in one model; sold per token via the OpenAI-compatible Liquid API and released as open weights under the revenue-capped LFM Open License.
Speech-to-speech 3
| Compare | Model | Price |
|---|---|---|
lfm3-audio-2bLFM3-Audio-2B |
$2.00 per 1M tokens | |
LiquidAI/LFM2.5-Audio-1.5B-JPLFM2.5-Audio-1.5B-JP |
– | |
lfm3-omni-2bLFM3-Omni-2B |
$2.00 per 1M tokens |