DROPGoogle launches Gemini 3.8 Flash TTS and Flash-Lite TTS///DROPGemini Omni brings 1080p generation to Google Vids///BREAKClaude Opus 5.5 is here: 1M context, $4 input / $20 output///BREAKOpenAI adds GPT-6 Sol and Luna to the GPT-6 family///PAPEROpus 5.5's system card documents gains—and evaluation limits///BREAKGrok 4.7 launches at $2 input / $6 output per million tokens///DROPGrok Voice Transcribe 2.0 targets difficult real-world audio///TRENDGoogle steers new Gemini API projects to 3.8 Flash or 3.5 Flash-Lite///DROPGoogle launches Gemini 3.8 Flash TTS and Flash-Lite TTS///DROPGemini Omni brings 1080p generation to Google Vids///BREAKClaude Opus 5.5 is here: 1M context, $4 input / $20 output///BREAKOpenAI adds GPT-6 Sol and Luna to the GPT-6 family///PAPEROpus 5.5's system card documents gains—and evaluation limits///BREAKGrok 4.7 launches at $2 input / $6 output per million tokens///DROPGrok Voice Transcribe 2.0 targets difficult real-world audio///TRENDGoogle steers new Gemini API projects to 3.8 Flash or 3.5 Flash-Lite///
DARC.DEV
All guides
GuideLevel · Start#Gemini#Voice#Multimodal

Gemini 3.8 is a family. Flash, Live, TTS and Cyber do different jobs.

A practical map of Google's September models, with real availability dates, introductory pricing and the distinction between generation and live dialogue.

DCDariusz Czechowicz23 Sept 20263 min readSource briefing
Gemini 3.8 is a family. Flash, Live, TTS and Cyber do different jobs.
Source-checked Sources & corrections
2 practical sections · checklist · sources

Source briefing. Provider claims are attributed to the sources below. This page does not claim a new independent DARC benchmark.

For: Curious readers and builders getting oriented

⚡ TL;DR — the 30-second version

  • Flash: multimodal inputs, text output, general-purpose workflows.
  • Live: real-time dialogue; TTS: generated speech from a script.
  • Flash Cyber has separate trusted-access rules.
  • Introductory Flash token pricing expires on 31 December 2026.

Gemini 3.8 Flash is Google's current general-purpose Flash release in the reviewed API documentation. It accepts text, images, video, audio and PDFs, but its output is text. The model page explicitly says the Live API, image generation and audio generation are not supported by this particular model.

Its input limit is 1,048,576 tokens and output limit is 65,536. Unlike a total-context window quoted by another provider, these are separate published limits; do not assume all million-token labels describe the same constraint. The current introductory token prices are $0.75 input and $3.75 output per million through 31 December 2026. Published standard rates starting January are higher.

“Do not route by the number 3.8. Route by the output you need and the interface the model actually supports.”
— Editorial interpretation

Gemini 3.8 Live and Live Extended Thinking were released on 15 September for audio-to-audio interactions. Asynchronous function calls let an agent continue a conversation while work runs in the background. The Extended Thinking variant is intended for additional background reasoning. Confirm product-specific rollout and session behavior before changing a voice application.

On 23 September Google introduced Gemini 3.8 Flash TTS and Flash-Lite TTS. These are text-to-speech generation models, with voice design and consent-verified replication. The announcement distinguishes features available at launch from voice remixing, which remains coming soon.

Flash Cyber is another distinct deployment. Access is offered to trusted defenders through the Fairwind Program; the general Flash API does not imply access to Cyber. Finally, Google says existing Gemini 2.5 users can continue on those models while new projects should use 3.8 Flash or 3.5 Flash-Lite. Restricted new access is not the same as retirement.

Field guide / 01

Choose the interface from the expected output

A document-to-text application, a live conversation and a scripted voiceover are three different request patterns. Write down input types, required output and turn-taking behavior before selecting the model name.

For live tools, plan how the user learns that a background action has actually completed. A model acknowledging a request while it continues speaking is useful, but it must not be confused with confirmation from the downstream service.

NeedRelevant familyDo not assume
Multimodal input, text response3.8 FlashNative speech output
Real-time voice dialogue3.8 LiveEvery text API option is supported
Scripted narration3.8 TTSAll voice features are universally available
Specialist cyber access3.8 Flash CyberPublic unrestricted access

Field guide / 02

Keep a small availability calendar

Track release date, your account's enabled features, preview status and price expiry separately. The documentation can be current while a particular workspace is still waiting for a staged rollout.

For introductory prices, schedule a budget review before the expiry date. Keep the published future rate in your forecast rather than silently assuming the promotional price is permanent.

Before you ship

0/5

Use this as a review aid. Checks reset when you leave this page.

Questions that come up

Can I use Flash as a drop-in replacement for Live?

No. Flash generates text from multimodal input; Live has a separate real-time conversation interface and behavior. Use the model-specific documentation.

Put it into practiceEvaluate an end-to-end voice agent

Sources & verification

Last checked 23 Sept 2026. Publication dates and access conditions belong to the linked originals.

  1. 01Gemini 3.8 Flash model specificationsGoogle↗
  2. 02Gemini 3.8 Flash — introductory pricing and migrationGoogle↗
  3. 03Gemini 3.8 Live and Live Extended ThinkingGoogle↗
  4. 04Gemini 3.8 text-to-speech says helloGoogle↗
  5. 05Gemini API release notesGoogle↗
Editorial policy and correction log →
#Gemini#Voice#Multimodal#Models