Google DeepMind added models/gemini-3.8-flash-lite-tts to the list of models it will answer with.

Google DeepMindMODEL-ROSTERADDEDWED SEP 23, 14:22

The line, as the page carries it

+models/gemini-3.8-flash-lite-tts: "version=3.8-flash-lite-tts, displayName=Gemini 3.8 Flash Lite TTS, inputTokenLimit=8192, outputTokenLimit=16384, supportedGenerationMethods=batchGenerateContent, countTokens, generateContent"

The record

DOCUMENThttps://generativelanguage.googleapis.com/v1beta/models?pageSize=200
SOURCEgdm-models-api
ROWmodels/gemini-3.8-flash-lite-tts
OPERATIONadded
BEFORENothing was recorded because the row did not exist.
AFTERversion=3.8-flash-lite-tts, displayName=Gemini 3.8 Flash Lite TTS, inputTokenLimit=8192, outputTokenLimit=16384, supportedGenerationMethods=batchGenerateContent, countTokens, generateContent
READ AT2026-09-23T14:22:59.822Z
SIGNALSmodelId

What went out

    When a lab changes a document

    The engine refetches every registered document on a schedule and files the line that moved. Model cards, system cards, pricing pages, API changelogs and deprecation notices all change without an announcement. Leave an address and Context Window writes when one does.