Gemini 3.8 Flash TTS returns 404 "Model not found" via API, despite working in AI Studio (Tier 1)

Gemini API
, , , ,
1

Hi Google Gemini team,

I’m experiencing an issue accessing Gemini 3.8 Flash TTS (gemini-3.8-flash-tts) through the Gemini API, despite the model being available and working correctly in Google AI Studio.

Environment:

  • Model: gemini-3.8-flash-tts

  • Voice: Bodi

  • SDK: google-genai 3.1.0 (Python)

  • Billing: Tier 1 (paid, prepaid)

  • API key: Created November 11, 2026, through Google AI Studio

  • API restrictions: Gemini API (generativelanguage.googleapis.com)

What works:

  • I can successfully generate speech using Gemini 3.8 Flash TTS with the Bodi voice inside Google AI Studio.

  • The model appears in my project’s rate-limit dashboard, with an allocation of 10 RPM, 10,000 TPM, and 100 RPD.

  • My API key is active and correctly restricted to the Gemini API.

What fails:

I’ve tested both API interfaces:

  1. client.models.generate_content_stream()

  2. client.interactions.create()

Both return the same error:

404 NOT_FOUND
Model 'gemini-3.8-flash-tts' not found.
Please verify the model name against the supported list:
https://ai.google.dev/gemini-api/docs/models

Additionally, calling client.models.list() and filtering for model names containing tts returns no results.

I’ve verified that the API key belongs to the intended Google Cloud project and that the Gemini API restriction is configured correctly.

Expected behavior:

The API should recognize gemini-3.8-flash-tts and allow speech generation using Bodi, consistent with the model availability shown in Google AI Studio and the official documentation.

Questions:

  1. Is Gemini 3.8 Flash TTS currently accessible through the public Gemini API for Tier 1 projects?

  2. Is there a difference between model availability in AI Studio’s speech-generation interface and availability through API keys?

  3. Does this model require additional activation, a particular API version, or a different endpoint?

  4. Is there a known issue causing this model to return 404 despite appearing in the project’s rate-limit dashboard?

I’m developing an automated long-form narration workflow and would appreciate clarification on the correct API configuration.

Thank you!

ModelTierLimist
ModelTierLimist2048×284 57.9 KB

2

Hello @dessislah ,

Yes, gemini-3.8-flash-tts is accessible through the public Gemini API for Tier 1 projects using your existing API key and the standard generativelanguage.googleapis.com endpoint, with no extra activation required. Seeing it in your rate-limit dashboard and working in Google AI Studio is expected.

The 404 happens because recent google-genai SDK 3.x releases default to API version v1, while audio output and TTS models (gemini-3.8-flash-tts) are currently available on v1beta only (and are omitted from models.list() on v1).

To fix this, configure the client to use v1beta—either in code or via the GOOGLE_GENAI_API_VERSION=v1beta environment variable:

from google import genai

client = genai.Client(http_options={"api_version": "v1beta"})

For more details, see Configure API version in an SDK and the Text-to-speech guide. Please try again with v1beta and let us know if you run into any issues.