brain/
← all entities
entitygenericartificial-intelligence

Lyria 3.5

Notes

Vintage: 2026-09. Primary evidence is Google's Sep 04, 2026 Keyword post plus official @GeminiApp X, as hydrated in 2026-09-04-lyria-3-5-lands-in-gemini-app-and-api-global. The model itself is older: Google launched Lyria 3.5 on Jul 29, 2026 in Flow Music. Sep 4 is a Gemini-app + Gemini API access expansion, not a new-model day. Fetched gemini.google/overview/music-generation/ still lists Lyria 3 / Lyria 3 Pro — treat that as lagging marketing HTML, not as a silent rename of 3.5 into 3 Pro. No Gemini API model id, price, or 3.5 length cap in the fetched text. Do not invent those. Neither Sep 4 X post returned a media attachment (x_video: false). gemini.google.com/music and deepmind.google/models/lyria were not fetched (timeout).

Lyria 3.5

One-line summary: Google's already-launched music-generation SKU (Jul 29 2026 in Flow Music) — on Sep 4 2026 Google said it is in the Gemini app and the Gemini API for all users globally on web and mobile. Not a Sep 4 model drop.

What it is

Google's Sep 04 Keyword post presents Lyria 3.5 as “our best-sounding music generation model” now available in the Gemini app and the Gemini API. That is issuer marketing, not an independent eval. The same post names Google Flow Music, Google AI Studio, and Google Vids as additional channels.

The Jul 29, 2026 Flow Music post is the model-intro date: musicality, lyrics, vocals/pronunciation, and tempo/duration control, rolling out that day in Flow Music. Prefer Jul 29 for the SKU's introduction. Prefer Sep 4 for Gemini-surface + API distribution of that already-named SKU.

This page records issuer-hydrated claims from the fetched Sep 4 blog, the fetched Jul 29 blog, the fetched music-overview page, the Feb 18 Lyria 3 Gemini-app post (recap-only), and official @GeminiApp X text. It does not re-file same-day omen-alpha, unsloth, gpt-6-astra overnight attach, gemini-3-8-flash, or weathernext-3.

Why it matters to this thread

Frontier model releases, deployment availability, and multimodality (audio / music generation) are in-scope. This is the first dated Lyria / Gemini-music page in the thread. File it as a product-access expansion, not as a foundation-model card.

Key facts (from 2026-09-04-lyria-3-5-lands-in-gemini-app-and-api-global)

Jul 29 2026 — model intro in Flow Music

Sep 4 2026 — Gemini app + API access expansion

  • From the same source (Create your best tracks yet with Lyria 3.5 in Gemini, Sep 04, 2026): “Lyria 3.5, our best-sounding music generation model, is now available in the Gemini app and the Gemini API,” with “more expressive vocals and richer musical arrangements” and “higher fidelity.”
  • From the same source (same blog, Gemini-app surface): select or describe genre and choose vocal or instrumental; new templates; flexibility to choose short or longer tracks.
  • From the same source (same blog): Lyria 3.5 “is available to all users globally on the web and in the mobile app,” and “also available for artists and AI creatives in Google Flow Music, and for developers and technologists through Google AI Studio, and Google Vids.”
  • From the same source (@GeminiApp, 2026-09-04): Lyria 3.5 is in Gemini; genre / vocal-or-instrumental / templates / short or longer tracks. No media attachment.
  • From the same source (@GeminiApp, 2026-09-04): all users globally on the web and in the app. The follow-up’s t.co target resolved to gemini.google.com/music (curl Location only; HTML fetch timed out). No media attachment.

Do not invent a Gemini API model id, a price, a rate limit, or a 3.5 length cap. The fetched Sep 4 blog names those channels and does not publish those numbers.

Overview page still sells Lyria 3 / Lyria 3 Pro (lagging marketing)

Fetched gemini.google/overview/music-generation/ does not name Lyria 3.5. Keep this as lagging marketing HTML, not as a contradiction of the dated Sep 4 blog, and not as a silent rename of 3.5 into 3 Pro.

  • From the same source (overview page): product still titled Lyria 3.
  • From the same source (overview page): SKU split is Lyria 3 (free, tracks up to 30 seconds) vs Lyria 3 Pro (higher limits with Google AI Plus/Pro/Ultra; tracks up to 3 minutes).
  • From the same source (overview FAQ): pick “Fast” for Lyria 3 / 30 seconds and “Thinking” or “Pro” for Lyria 3 Pro / up to 3 minutes.
  • From the same source (overview page): 18+ availability wherever the Gemini app is available; SynthID watermarking.

Do not collapse 3.5 into 3 Pro. Do not stamp the 30-second / 3-minute caps onto Lyria 3.5.

Feb 18 2026 Lyria 3 Gemini-app launch (recap-only)

  • From the same source (Use Lyria 3 to create music tracks in the Gemini app, Feb 18, 2026): prior Gemini-app launch — 30-second tracks, 18+, listed languages, paid-plan higher limits. Its related-story chip now points at the Sep 4 Lyria 3.5 Gemini post. Recap-only — do not mint a Lyria 3 entity from this chip.

What this source does not establish

  • No API grain. The blog names the Gemini API, AI Studio, and Vids. Fetched text has no model id, price, or rate limit. Do not invent those.
  • No 3.5 length cap. Sep 4 blog/X say “short or longer tracks.” The 30-second / 3-minute numbers belong to the overview page’s Lyria 3 / Lyria 3 Pro split. Do not invent a 3.5 cap.
  • No fetched music URL or DeepMind model page. gemini.google.com/music timed out after curl Location. deepmind.google/models/lyria timed out. Do not cite either as a retrieved source.
  • No independent eval. “Best-sounding” is Google marketing.
  • Flow Music vs Gemini stack is not evidenced. Jul 29 launched 3.5 in Flow Music; Sep 4 adds Gemini app/API and restates Flow Music.
  • Quiet labs after 10am (@karpathy, @OpenAI, @AnthropicAI, @GoogleDeepMind, @sama, @Zai_org, @opencode) returned zero original posts in the clip’s window — a negative finding, not grain.
  • News-cluster summaries (Astra recaps, Amodei-meme chatter) were pointers only and are not cited.
  • Not a re-file of gemini-3-8-flash, weathernext-3, gpt-6-astra, omen-alpha, or unsloth.
  • Adjacent Sep 15 Live / audio suite pointer: 2026-09-15-google-gemini-3-8-live-3-8-live-extended-thinking restates Lyria 3.5 as an already-filed suite item on the developer companion — not a Sep 15 music launch and not a rewrite of this page. Live SKUs live on gemini-3-8-live.
  • No ticker, 8-K, or stock-market tag.

Contradictions / tensions

  • SKU rename vs page lag: Sep 4 blog/X say Lyria 3.5 for all Gemini users globally, short or longer tracks. The overview page still sells Lyria 3 / Lyria 3 Pro with 30s vs 3-minute and paid-plan language. Treat the overview as lagging marketing HTML. Do not collapse 3.5 into 3 Pro. Not reconciled.
  • Access date vs intro date: Jul 29 is the model-intro in Flow Music; Sep 4 is Gemini-surface + API distribution. Do not file Sep 4 as a new-model day.
  • “Best-sounding” is issuer rhetoric, not a third-party bake-off.
  • Named API / Studio / Vids channels vs missing SKU card: surfaces are named; model id / price / rate limit / 3.5 length cap are not.

Open questions

  • What Gemini API model id, price, and rate limit apply to Lyria 3.5? The fetched Sep 4 blog does not publish them.
  • What length cap, if any, applies to Lyria 3.5 itself? Do not borrow the overview page’s 30s / 3-minute Lyria 3 / 3 Pro numbers.
  • Do Flow Music and Gemini share one stack? Not evidenced here.
  • When (if ever) is gemini.google/overview/music-generation/ rewritten to name Lyria 3.5?

Sources

Related

Referenced by
Entities
Gemini 3.8 Flash and 3.8 Flash CyberGemini 3.8 Live and 3.8 Live Extended ThinkingRumik AIrumik-oss 1