LFM2.5-Audio-1.5B
LFM2.5-Audio-1.5B is an English speech-and-text checkpoint that remains downloadable after deprecation.
Deprecated API; check access details
What the release provides
It combines an LFM2.5 backbone, a FastConformer audio encoder and a speech-token generator. Interleaved output supports conversation; sequential generation supports transcription or synthesis.LFM2.5-Audio-1.5B official model card
How to use it well
Practical guidance. For an existing integration, preserve its audio preparation and generation mode while comparing the replacement. Evaluate the full conversation, including interruption handling and spoken output, rather than replacing the model from a transcript-only comparison.
Downloadable but no longer maintained
Liquid marks this model deprecated. Weights remain downloadable, but the release no longer receives updates.Liquid model deprecations
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Deprecated API; check access details |
| Access | Downloadable |
| Model identifiers | LiquidAI/LFM2.5-Audio-1.5B |