Is Gemini (Google) Down?

GEMINI 3.8 FLASH AND 3.8 FLASH CYBER -- THE THIRD FLASH RELEASE IN SIX WEEKS, AND THE INTRO PRICE DID NOT RESET (2026-09-02, vendor-primary): Google shipped **Gemini 3.8 Flash** and **Gemini 3.8 Flash Cyber**, describing 3.8 as 'our best reasoning and coding model yet, at the same speed and low cost of 3.7' and noting this is its **third Flash release in only six weeks**, three weeks after 3.7 Flash. **PRICING -- AND THE TRAP IS THE SAME ONE 3.7 FLASH CARRIED, ON THE SAME DEADLINE:** 3.8 Flash is '**available at the same introductory price as 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens**'. The footnote is explicit: '**Introductory price expires on December 31, 2026. Starting January 1, 2027, $1.50/1M input tokens and $7.50/1M output tokens will apply.**' **So the new model inherits the old model's expiry date rather than getting a fresh twelve months -- the discount window is now four months shorter than it was for anyone who adopted 3.7 Flash in August.** Budget on the January 1 doubling, not on today's rate. **BENCHMARKS, VENDOR-PUBLISHED:** **54.9% on HLE-Verified**; Google says 3.8 Flash 'outperforms most larger frontier models' on **DeepSWE v1.1** long-horizon software engineering 'at a fraction of the cost', and beats 3.7 Flash and other frontier models on **Vals Finance Agent V2** and **Harvey's Legal Agent Benchmark**. **AN HONEST COST CAVEAT GOOGLE PUBLISHES AND AGGREGATORS WILL DROP:** the gains come from the model '**working harder**' -- 'executing extra reasoning steps, and calling tools iteratively', and 'at times, the model might use more tokens to maximize performance, especially at higher effort levels'. **Same per-token price, potentially more tokens per task -- so the effective cost per task can rise even though the rate card did not move.** Google explicitly tells efficiency-first users to lower the effort level or **stay on 3.7 Flash, which 'remains fully supported'** -- so this is not a forced migration. **GEMINI 3.8 FLASH CYBER IS NOT PURCHASABLE:** it is available 'to a select set of trusted defenders' via the new **Fairwind Program**, aimed at trusted government authorities, critical-infrastructure operators and software maintainers, with prioritized access by application. Google reports frontier CyberGym performance at **2.3-5.2x lower cost than other leading frontier models**, and says its Cloud Vulnerability Research team used it to find a critical foundational vulnerability **in under two hours** where discovery normally takes months. **AVAILABILITY OF 3.8 FLASH (the one you can actually use):** developers via Google Antigravity, the Gemini API through Google AI Studio, Android Studio and Stitch; enterprises via Gemini Enterprise; consumers on **Google AI Pro and Ultra** across the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets. **It reached GitHub Copilot the next day, 2026-09-03**, for Pro, Pro+, Max, Business and Enterprise.

GEMINI 3.5 TRANSCRIBE -- NEW SPEECH-TO-TEXT MODEL, AND IT RETIRES CHIRP 3 AS THE REFERENCE POINT (2026-08-26, vendor-primary): Google launched Gemini 3.5 Transcribe, its '**most precise speech-to-text model yet**', in the Gemini API via Google AI Studio and the Gemini Enterprise Agent Platform. **Two separate APIs, two model IDs:** real-time streaming with sub-second latency via the Live API using `gemini-3.5-transcribe-live`, and pre-recorded processing with speaker attribution and word-level timestamps via the Interactions API using `gemini-3.5-transcribe`. **Vendor-cited accuracy, attributed to Artificial Analysis (third party, not Google's own harness): average WER 4.0% streaming / 2.6% non-streaming**, and on the multilingual FLEURS benchmark **5.50% streaming / 5.04% non-streaming**. Time-to-final-transcription improves **70% versus Chirp 3**. **85+ languages** auto-detected; **multi-speaker attribution is capped at three speakers, with 3+ explicitly experimental** -- that ceiling is the practical limit for meeting transcription and no aggregator will mention it. Also does disfluency cleanup (self-corrections, filler removal, auto-formatting), custom vocabulary, and **function calling to delegate to other Gemini models -- currently only in the Gemini macOS app**. **NO PUBLISHED RATE CARD in the launch post.**

GEMINI OMNI 1.1 FLASH -- GENERATIVE VIDEO GETS PRODUCTION CONTROLS (2026-08-27, vendor-primary): Google shipped Gemini Omni 1.1 Flash, framed explicitly as making Omni 1.1 '**production-ready for professional use**' via the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform. **Four changes, all about control rather than raw quality:** (1) **Scene extension** -- continue generating from where an existing clip left off; the model now analyses **up to 10 seconds of prior context, versus previous models that referenced only the final second**, and you can extend in 10-second increments to a **cumulative 40 seconds**. (2) **First and last frame interpolation** -- set start and end frames to control camera moves and transitions. (3) **360p preview renders** at lower credit cost to iterate before committing to a full render. (4) **4K upscaling** for final output. **READ THE 40-SECOND CEILING AS THE REAL LIMIT:** this is a cumulative cap, not a per-call cap, so Omni 1.1 is still a shot-generator, not a long-form video tool. **NO PRICING WAS PUBLISHED for the 1.1 tier in the announcement** -- do not carry Omni Flash's $0.10/sec across without confirming it on a rate card.

A FREE YEAR OF GOOGLE AI FOR STUDENTS -- AND THE AUTO-RENEW TERM IS THE PART THAT WILL CATCH PEOPLE (2026-08-19, vendor-primary): Google opened a back-to-school offer that is materially the largest giveaway on this page. **US college students get 12 months of Google AI Pro free**, which Google itself values at **$19.99/mo** -- unlocking **4x higher usage limits in Gemini**, access to **Gemini Spark**, Gemini in Gmail and Docs, **5 TB of storage** and Google Health Premium. **Outside the US, eligible students get 12 months of Google AI Plus free** instead: **Gemini Omni**, 2x higher usage limits, 400 GB storage. There is also a **Google AI Pro + YouTube Premium student bundle at up to 70% off**. **THE TERMS, READ OFF GOOGLE'S OWN PLANS PAGE RATHER THAN THE BLOG POST:** eligibility requires **verification via SheerID at sign-up**; student offers are '**valid for up to 4 years and may automatically renew at the standard monthly price once student status concludes**'; they are '**not available on school-issued accounts**' and are individual-use only. **That auto-renew-to-full-price clause is the single most important sentence for a reader** and it does not appear in the announcement blog post -- it is in the plans-page footnotes. Anyone claiming this should set a calendar reminder for month 12. **SHIPPED ALONGSIDE:** a dedicated **student hub at gemini.google.com/students** collecting study notebooks, flashcards and practice quizzes; study notebooks (launched June) gain **graphs and images 'in the coming weeks'** and can write exam and assignment dates from a syllabus into Google Calendar with permission; interactive 3D visualizations; and Gemini Live for talking through Drive materials. **COMPETITIVE READ:** this lands the day after OpenAI shipped ChatGPT for Teens and its CodeAI education partnership. Both labs moved on the education market in the same 48 hours, but the shapes differ -- **OpenAI restricted a tier (automatic placement, safety-first), Google discounted one (free access, feature-first)**

GEMINI IN CHROME REACHES ALL US ANDROID USERS, AND AUTO BROWSE ARRIVES ON MOBILE FOR PAID TIERS (2026-08-18, vendor-primary): Google made **Gemini in Chrome available to all Android users in the U.S.** -- summarising long articles, answering questions about a page, and connecting to Google apps including Calendar and Keep without tab-switching, plus **Nano Banana image generation and customisation in-browser**. Reached via the Gemini icon or the three-dot menu on the top bar. **THE TIER SPLIT IS THE FACT THAT MATTERS:** the assistant is free to all US Android users, but '**AI Pro and AI Ultra subscribers on Android in the U.S.**' additionally get **auto browse**, Chrome's agentic mode -- booking parking, updating a recurring order, organising travel. So the browsing assistant is free; **the agent is paywalled**, and only in the US, and only on Android. **SECURITY POSTURE, VENDOR-STATED:** models are 'trained to detect known threats, such as **prompt injection**', and auto browse '**is designed to ask for confirmation before completing some sensitive tasks**' -- note the hedge in 'some', which Google does not enumerate. **HOW THIS RELATES TO THE 7/30 SPARK ENTRY BELOW:** that update brought Chrome auto-browse to the AI Pro tier and expanded Spark to 160 countries on desktop; this is the Android leg of the same push, and it is **US-only at launch** rather than inheriting that 160-country footprint -- do not assume the desktop geography applies here

NEW FLAGSHIP WORKHORSE -- GEMINI 3.7 FLASH (2026-08-13, vendor post): Google shipped **Gemini 3.7 Flash**, billed as 'our most intelligent workhorse model yet for coding and agents' -- and it **supersedes Gemini 3.6 Flash after only three weeks** (3.6 Flash was 7/21). **Vendor-published benchmarks, all versus 3.6 Flash:** FrontierCode 1.1 Main **43.6% (from 34.4%)**, DeepSWE v1.1 **65.3% (from 49.0%)**, WebDev Arena Elo **1588 (from 1538)**, GDP.pdf **34.0% (from 22.0%)**, AutomationBench **30.4% (from 17.0%)**. These are first-party figures; third-party verification pending. **PRICING IS INTRODUCTORY AND IT DOUBLES -- read this before you build on it:** **$0.75/M input and $3.75/M output through 2026-12-31**, then **$1.50/M input and $7.50/M output from 2027-01-01** (confirmed on both the launch post and ai.google.dev's pricing table). Note what that means in context: the post-promo price is **exactly Gemini 3.6 Flash's price**, so the intro rate is a 50% launch discount rather than a durable price cut -- budget on $1.50/$7.50 for anything running past New Year. **Availability day one:** Google AI Studio and Android Studio, the Gemini API, Google Antigravity, the Gemini Enterprise Agent Platform, and **Gemini Spark in the Gemini app for Pro and Ultra subscribers** -- plus, confirmed since, **GitHub Copilot on the same day (2026-08-13)**, across Copilot Pro, Pro+, Max, Business and Enterprise, though Business/Enterprise admins 'must enable the Gemini 3.7 Flash Preview policy in Copilot settings' first (note GitHub carries it under a **Preview** policy label, so do not assume Copilot's 8/26 default-enablement flip switches it on automatically). **The strategic read:** Google has now shipped three Flash generations (3.5 Flash-Lite, 3.6, 3.7) in roughly seven weeks while **Gemini 3.5 Pro remains unshipped and partner-only** -- the Flash line is where Google's coding-model progress is actually landing, and buyers waiting on a Pro-tier answer should plan around Flash instead