Hi team,
Following the pattern other developers have used here for image-preview allowlist requests, I’d like to request increased access to **`gemini-3.1-flash-tts-preview`** on Vertex AI for our production application.
### Project details
- **GCP Project ID:** `gen-lang-client-05462384…
- **Region:** `us-central1`
- **Billing:** active billing account linked (verified — Cloud Functions deployments and Cloud Billing Kill Switch budget alerts both working)
- **SDK:** `@google/genai` with `vertexai: true`
- **Existing quota approval (precedent):** Case `71341105` — `gemini-2.5-pro-tts` was approved at 100 RPM on May 21, 2026
### Model requested
`gemini-3.1-flash-tts-preview` — currently returning frequent 429 / 503 quota-exhausted errors even after waiting 30+ minutes between calls, suggesting Dynamic Shared Quota (DSQ) pool exhaustion rather than per-project rate limits.
### Use case
OptiRunner is a Dutch AI running coach app for Android (Google Play Store, project Valideas B.V., KvK 68232705). The app uses Vertex AI Gemini TTS to provide real-time Dutch-language coaching with the Autonoe voice during outdoor running sessions. Each ~30-45 minute training session generates 7-10 TTS calls.
Specific reasons we need 3.1-flash-tts-preview access:
1. **Voice quality:** subjective testing shows 3.1 Flash TTS produces noticeably warmer, more natural Dutch coaching delivery vs. 2.5 Pro TTS
2. **MP3 fallback bank rebuild:** we maintain a bank of ~365 pre-rendered Dutch coaching clips that we want to upgrade to 3.1 quality. We’re currently blocked at 63/365 clips due to DSQ exhaustion
3. **Lower latency** vs. 2.5 Pro is critical for real-time outdoor coaching delivery
### Volume estimate
- **Current beta:** 12 testers (June 2026 closed beta on Google Play)
- **Soft launch:** ~100 paying users (Q3 2026)
- **First year target:** ~1,000-2,000 paying users (Netherlands only, Android only)
Estimated daily TTS calls at full scale: 5,000-10,000 calls/day during peak hours (morning + evening run times in CET timezone).
### What we’re asking
Allowlist access to `gemini-3.1-flash-tts-preview` for project `gen-lang-client-0546238495` at a sustained rate sufficient to support our scale projections. We’re flexible on regional vs. global endpoint — whatever works best for the DSQ pool architecture.
### Why not the standard support route
Case `71341751` was filed via Cloud Support on May 17, 2026. Support engineer Ness confirmed on May 23 that Flash Preview TTS requires an internal business approval or assigned Account Executive that support engineers cannot themselves provide. Following his suggestion, we’re now requesting through this community/forum channel which other developers have used successfully for similar preview-model allowlist requests (e.g., `gemini-3.1-flash-image-preview`).
We have a Pro 2.5 TTS fallback active in production right now (Pro 2.5 quota at 100 RPM, case 71341105), so this is not blocking our beta launch. We’re requesting 3.1 Flash TTS access to enable our quality and latency upgrade path.
Happy to provide any additional information needed.
Thanks,
Andreas de Antoni
Valideas B.V. — Founder & Sales Contact