$ cat wiki/models/lyria-3-5.md
Lyria 3.5
modelupdated 2026-07-30created 2026-07-30
Spec
| Attribute | Value |
|---|---|
| Developer | Google DeepMind |
| Released | 2026-07-29 |
| Announced | 2026-07-29 |
| Context window | unknown |
| Pricing | unknown |
| License | unknown |
| Availability | Google Flow Music (rolling out) |
| Announced and rolled out on the same day; no separate preview period was stated | |
| (source). |
Release Date
2026-07-29, through Google Flow Music, Google's AI music creation platform. Predecessor Lyria 3 was introduced roughly a month earlier (deepmind.google/models/lyria).
Benchmarks
None published. The announcement describes improvements qualitatively and gives no scores, listening-test results or comparisons against Lyria 3 or any third-party model (source).
Stated advances, in the announcement's own terms:
| Dimension | Claim |
|---|---|
| Musicality | "richer, more complex melodic structures that sound more natural" |
| Lyrics | "higher quality lyrics with improved prompt adherence and structural awareness" |
| Vocals | "more realistic and emotionally nuanced vocals", improved pronunciation |
| Creative control | easier control over tempo and duration of outputs |
Use Cases
Music creation inside Google Flow Music. Platform capabilities named in coverage: editing specific sections of a track, rewriting lyrics, translating lyrics into another language, and changing other aspects of generated music (source).
Compared To
| Model | Developer | Note |
|---|---|---|
| Lyria 3 | Google DeepMind | direct predecessor, ~one month earlier; no published comparison |
| Muse Video | Meta AI | generative video with native audio — adjacent, not directly comparable |
| Gemini Omni | Google DeepMind | the general multimodal line, of which Lyria is the music-specific branch |
| No head-to-head figures exist for any of these pairs. |