Gemini Just Got a Big Update with Gemini 3.7 Flash— Here’s Everything New.

Image : Gemini 3.7 Flash India: What’s Live, What’s Not ( Real Verified)

---

Google has pushed out one of its busiest update windows for Gemini in recent months. Over the last three days, a new model rolled out by default, a long-requested watermark control appeared in settings, Sheets got a Gemini-powered “mini-app” builder, and Google confirmed a wave of new connected apps. Here’s what actually changed — including what I’ve verified directly on my own account.

---

Gemini 3.7 Flash Is Rolling Out — Tied to AI Pro, Not Yet on Free Tier

The most immediate change: Gemini 3.7 Flash has started replacing 3.6 Flash as the default option in the Gemini model picker, tagged “New” alongside 3.5 Flash-Lite and 3.1 Pro.

I checked this across every smartphone in my house, each signed into a different Google account. Every account on a Google AI Pro plan (including Pro activated via Jio’s bundled offer) showed 3.7 Flash live and set as default. One device on the free Gemini tier was still showing the older model — no 3.7 Flash yet.

This matches what Google has confirmed officially: in the Gemini app, 3.7 Flash currently reaches consumers only through Gemini Spark, Google’s 24/7 personal AI agent — and Spark itself requires an AI Pro or Ultra subscription. There is no free-tier route to 3.7 Flash in the app right now. If you’re on the free plan, you won’t see it yet — not because the rollout hasn’t reached you specifically, but because free accounts don’t have Spark access at all.

Note: some outlets (Ars Technica, felloai.com) reported that 3.7 Flash is only accessible through Spark and is not selectable in the regular Gemini chat interface. My own screenshot shows 3.7 Flash directly selectable and set as default in the standard model picker — not gated behind opening Spark separately. This may reflect the rollout reaching different surfaces at different times, or regional differences in how the UI is structured. I’m going with what I can directly verify on my own account.

According to Google, this release arrives just three weeks after 3.6 Flash and focuses on:

• Coding and agent workflows
• Software engineering tasks
• Web development
• Complex document analysis

Pricing: Google is offering 3.7 Flash at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens through the Gemini API — half the launch price of 3.6 Flash. This pricing holds through December 31, 2026, after which it rises to $1.50 / $7.50 per million tokens.

The model also now powers Gemini Spark, Google’s 24/7 personal AI agent (available to AI Pro and Ultra subscribers), with Google claiming improved tool use across Google Workspace apps.

Note: I haven’t yet run my own coding/agent benchmarks against 3.7 Flash — this section reflects Google’s stated improvements. I’ll update with first-hand comparisons once I’ve tested it directly.

What Actually Improved: 3.7 Flash vs. 3.6 Flash (Google’s Own Numbers)

Google published specific benchmark scores comparing 3.7 Flash to its predecessor. These are Google’s self-reported figures, not independently verified by me, but they’re worth knowing since “substantial improvements” is vague on its own:

Benchmark	What it measures	3.6 Flash	3.7 Flash
FrontierCode 1.1 Main	Production code quality	34.4%	43.6%
DeepSWE v1.1	Long-horizon software engineering	~49%	65.3%
Terminal-Bench 2.1	Agentic terminal coding	78.0%	85.8%
Code Arena (WebDev)	Web development, Elo score	1538	1588
AutomationBench	Business workflow automation	17%	30.4%
GDPVal-AA	Real-world knowledge work tasks	22%	34%

Some outlets report the 3.6 Flash DeepSWE baseline as 48.6% rather than 49.0% — a minor discrepancy across sources, but the roughly 16-17 point jump to 65.3% holds either way.

For context, Google also compared 3.7 Flash against competing models on the same benchmarks: it edges ahead of Claude Sonnet 5 and GPT-5.6 Terra on FrontierCode (43.6% vs. 42.7% and 41.3%) and on Code Arena web development, but trails GPT-5.6 Terra on DeepSWE (65.3% vs. 69.6%) and Terminal-Bench (85.8% vs. 87.4%). It’s a genuine step up from 3.6 Flash, not a clean win across every benchmark against rivals.

Who Actually Benefits From This Update

Given where the gains are concentrated, this update matters most to a specific set of users rather than everyone equally. Developers and freelancers who use Gemini for coding, debugging, or building web layouts get the clearest upgrade — the FrontierCode and Terminal-Bench gains translate directly into fewer failed attempts and less manual cleanup. Anyone running AI agents for repetitive business workflows (data entry, document processing, status updates) should also notice a real difference, given the jump on AutomationBench.

Google also claims improved reasoning in specialized domains like law, finance, and biochemistry, so professionals using Gemini for research or domain-specific analysis in those fields may see better output too. For everyday conversational use — quick questions, writing help, general summarization — the improvement is likely to be far less noticeable, since the biggest gains are concentrated in coding and agentic tasks rather than general chat quality. And since access currently runs through Spark, Chrome, or AI Mode on an AI Pro/Ultra plan, free-tier users in India don’t benefit from any of this yet.

---

Gemini 3.7 Flash Also Arrives in Google Search’s AI Mode and Chrome

A day after the app launch, Google added Gemini 3.7 Flash as a selectable model inside AI Mode in Google Search — confirmed by Robby Stein, VP of Product for Google Search, on August 14.

• You can pick it from the model menu in AI Mode’s “Ask anything” bar (tap the “+” icon), listed under “Gemini 3 models” next to “Auto” and “Pro”
• Google says it’s “better at following instructions + understanding your intent”
• Like the app rollout, this is gated to AI Pro and Ultra subscribers — free-tier AI Mode users don’t get the option
• Google hasn’t said whether 3.7 Flash will become AI Mode’s default model or factor into its automatic model routing

I checked this myself in Chrome’s built-in Gemini side panel, and 3.7 Flash is there too — listed alongside Auto, 3.5 Flash-Lite, 3.1 Pro, and Extended thinking, with 3.7 Flash already selected. So this isn’t limited to the Gemini app or Search’s AI Mode specifically — it’s showing up wherever Gemini surfaces inside Google’s own products for AI Pro accounts.

For most readers this is a minor addition — it gives you another model to manually compare against Auto/Pro for a given search, rather than changing what you get by default.

---

Also Read : Ask Google Pay: Google Pay’s New AI Assistant Explained (How to Enable It, What It Can and Can’t Do)

Visible AI Watermarks Can Now Be Turned Off (Rolling Out — Not Yet on My Account)

Google announced a new “Media watermark” toggle in Gemini settings that lets users remove the visible “sparkle” watermark that normally appears in the corner of AI-generated images (Nano Banana), video (Omni), and music (Lyria).

Important distinction: this only removes the visible mark. Google says invisible SynthID and C2PA metadata will still be embedded in all AI-generated content regardless of the toggle — so Gemini or Google Search can still detect and confirm AI-generated media even with the visible watermark off.

I checked my own account on August 15, 2026, and this toggle has not landed yet in India. Once it does, the setting will reportedly live under Settings → Media Watermark. I’ll publish a dedicated how-to with real screenshots the moment it appears.

One caveat from Google: the toggle won’t be available in countries where local law requires visible AI labeling on generated content — worth watching for India specifically once it rolls out further.

---

Sheets Canvas: Turn Spreadsheets Into Mini-Apps

Google also introduced Sheets canvas, a new Gemini-powered feature in Google Sheets that converts rows and columns into interactive dashboards, trackers, or seating charts using a natural-language prompt — no formulas required.

Key details:

• Type a prompt like “create a visual study tracker” and Gemini builds the interface
• Changes sync in both directions — edit the canvas or the underlying sheet, and both update in real time
• You can keep refining it by prompting Gemini for layout, design, or functionality changes
• Available in English for Google AI Pro and Ultra personal accounts, and rolling out to select Workspace Business/Enterprise Standard/Plus editions and AI Pro for Education

I plan to test this directly in a live Sheet and document the actual output — this section will be updated with real screenshots once I’ve built something with it.

---

New Connected Apps Rolling Out to Gemini

Google confirmed a new batch of third-party app integrations coming to Gemini “over the next few weeks,” organized into four categories:

Productivity & creativity: Granola, Otter.ai, Wix Local & entertainment: Fever, GetYourGuide, Localiza, OpenTable (UK), Ticketmaster Music: iHeartRadio, Pandora Home, health & lifestyle: Angi, Thumbtack, Zocdoc

Worth flagging for Indian readers: several of these (Ticketmaster, Angi, Zocdoc, Localiza) are primarily US/UK-focused services with limited or no presence in India, so the practical value of this batch will vary a lot depending on your location. These join existing connected apps like Canva, GitHub, Google Home, Instacart, and WhatsApp.

---

Rollout Status at a Glance (India, as of August 15, 2026)

Update	Status in India
Gemini 3.7 Flash (Gemini app)	Live on AI Pro accounts I tested; not seen yet on a free-tier account
Gemini 3.7 Flash (Chrome Gemini panel / AI Mode)	Confirmed live on my own AI Pro account in Chrome
Gemini Spark on 3.7 Flash	Live for AI Pro/Ultra subscribers per Google
Media watermark toggle	Not yet on my account
Sheets canvas	Not yet independently tested; Google states it’s live for AI Pro/Ultra
New connected apps	Rolling out over coming weeks; several are US/UK-only services

---

What This Means If You Use Gemini in India

• 3.7 Flash is live on AI Pro accounts — confirmed across multiple devices and separate Google accounts in my own household, including Pro activated via Jio. One free-tier account I checked has not received it yet, so this currently looks like a paid-plan rollout in India.
• Chrome and AI Mode also show 3.7 Flash — I confirmed this directly in Chrome’s Gemini side panel on my own AI Pro account, same Pro/Ultra gating as the app.
• The watermark toggle hasn’t reached my account yet — don’t be surprised if Settings still looks the same on yours.
• Sheets canvas is the most immediately useful update for small business owners and freelancers already on Workspace — no separate rollout wait if you’re on a supported plan.
• The new connected apps are largely US/UK-centric for now; don’t expect most of them to be relevant day one if you’re based in India.

I’ll keep this article updated as each of these features actually lands on my account, with real screenshots replacing the placeholders above.

---

Sources: Google’s official blog (blog.google), Google DeepMind model card, Reuters— cross-checked against my own Gemini accounts as of August 15, 2026.

---