Gemini 3.7 Flash 'Half Price'? What the Official Pricing Page Says About When It Ends
- Free tier: Yes
- Cheapest paid plan: US$4.99/mo and up
- Free quota: Free for anyone with a Google account. Model available: Gemini 3.6 …
- Last checked: 2026-11-19
Article last updated: 2026-08-14
On 2026-08-13, Google announced Gemini 3.7 Flash, and the official blog post states in black and white: “introductory price of half the original 3.6 Flash cost” — sounds like a big price cut.
But when I clicked through to Google’s own API pricing page, here’s what it actually says: “US$0.75 through December 31, 2026. US$1.50 starting January 1, 2027.”
In plain terms: this isn’t a price cut, it’s a promo running until the end of the year. And on that same page, the older model 3.6 Flash — which was only announced last month — is now listed at the exact same price. Below, I lay out the numbers, plus three things you can do today.
I. What Happened, in 30 Seconds
- On 2026-08-13, Google announced Gemini 3.7 Flash, just three weeks after its predecessor 3.6 Flash (2026-07-21).
- Positioned around “coding and agents (AI that automatically does tasks for you),” the official blog lists these benchmark gains: FrontierCode 1.1 Main improved from 34.4% to 43.6%, DeepSWE v1.1 from 49.0% to 65.3%, document-reading GDP.pdf from 22.0% to 34.0%, and AutomationBench from 17.0% to 30.4%.
- Specs: a 1 million token context (how much content can be fed in at once), with a maximum output of 64,000 tokens.
- The same day, GitHub also announced that Copilot users can now select this model.
(Verification date for the above: 2026-08-14, sources at the end of the article)
II. This Site’s Investigation: “Half Price” Compared to What?
This is the crux of the article. Here are both generations’ prices side by side (per million tokens, USD):
| Item | Input | Output | Period |
|---|---|---|---|
| 3.6 Flash list price on launch day (official blog, 2026-07-21) | US$1.50 | US$7.50 | At the time |
| 3.7 Flash current promo price | US$0.75 | US$3.75 | Until 2026-12-31 |
| 3.7 Flash price from 2027 onward | US$1.50 | US$7.50 | From 2027-01-01 |
| 3.7 Flash Batch promo price | US$0.375 | US$1.875 | Until 2026-12-31 |
Price verification date: 2026-08-14 (read directly from Google’s official API pricing page and official blog)
See it now? The so-called “half price” is comparing the promo price against the list price at the time the previous generation launched. Once January 1, 2027 arrives, Gemini 3.7 Flash’s price will return to US$1.50 / US$7.50 — exactly what 3.6 Flash was priced at when it launched last month, down to the cent.
This is exactly the old trap this site has fallen into six times before: a promo price is not a list price. If you budget for next year based on the promo price, next year you’ll be in for a shock.
There’s Something Even More Interesting: 3.6 Flash Is Now the Same Price
I went through the 3.6 Flash section on the same page line by line (read twice on the same day to confirm): 3.6 Flash is now also listed as “US$0.75 through December 31, 2026. US$1.50 starting January 1, 2027.”
In other words, right now, whether you use 3.6 or 3.7, the cost per million tokens is identical. Since the price is the same and 3.7 scores higher, there’s not much reason to stick with 3.6 — but this also means the claim “switching to 3.7 saves you half” doesn’t actually hold today, since staying on 3.6 costs the same.
⚠️ Per this site’s rule 9.2: the point that “3.6 Flash is currently listed at the same price” is currently supported by only a single source (Google’s own official pricing page). This site read it twice on the same day (2026-08-14) to confirm, but could not find a second independent source to cross-verify. Defer to the official announcement.
III. At a Scale You Can Feel
How much is “one million tokens”? Google’s official documentation states: “a token is equivalent to about 4 characters. 100 tokens is equal to about 60-80 English words.” Working backward, 1 million tokens is roughly 4 million characters, or 600,000–800,000 English words — about the length of a thick novel.
This site’s estimate (a hypothetical, not official data): a small team consuming 30 million input + 5 million output tokens per month —
- Now: US$41.25/month
- From 2027-01-01: US$82.50/month
- That’s US$41.25 more per month, or US$495 more per year
Want to bring that down? The official pricing page states that Batch mode is half the standard price — US$0.375 / US$1.875 during the promo period. The tradeoff is no real-time response, which suits “submit today, get results tomorrow” bulk work — such as processing thousands of customer service records or batch translation.
IV. What Most Readers Want to Know: Can I Use the Free Tier?
A. Mobile/web Gemini App → Currently, no access to 3.7 Flash.
This site read directly from gemini.google/subscriptions on 2026-08-14, and the model listed under the free plan is still 3.6 Flash, with an additional note: “Varying access to 3.1 Pro” (access to Pro fluctuates). 3.7 Flash on the App side runs through Gemini Spark, and the official blog explicitly states it’s for Google AI Pro/Ultra subscribers, across 160+ countries.
Subscription list prices confirmed the same day: AI Plus US$4.99 / AI Pro US$19.99 / AI Ultra starting at US$99.99 (up to US$199.99), all monthly list prices.
B. Want to try 3.7 Flash for free → Go through Google AI Studio. On the official pricing page, the free tier column for 3.7 Flash reads “Free of charge,” so developers can access it at no cost. However, the official rate limits page does not list numbers (RPD/RPM) for 3.7 Flash’s free tier — it just tells you to check your actual quota in the AI Studio dashboard. Per this site’s rule 1.2: if it can’t be found, we say so — we don’t estimate.
C. GitHub Copilot Free → Not available. GitHub’s update announcement from 2026-08-13 states it’s rolling out to Copilot Pro, Pro+, Max, Business, and Enterprise, with Copilot Free not on the list. Business/Enterprise admins also need to enable the “Gemini 3.7 Flash Preview” policy in the admin dashboard before teammates can select it. The announcement does not specify this model’s premium request multiplier, and this site will not fill that in on its behalf.
V. What Couldn’t Be Verified, I’m Telling You Too
- Whether enterprise (Vertex AI/Google Cloud) pricing follows the same promo: On 2026-08-14, this site attempted to fetch Google Cloud’s generative AI pricing page twice; the content returned was truncated, so this could not be confirmed — we won’t guess. If you use Google Cloud, please verify directly in your own dashboard.
- Whether the promo will be extended: Google only states it runs through 2026-12-31; nothing is said about afterward. All planning should currently assume the price will revert to US$1.50/US$7.50.
VI. The Following Is Ask Cat Editorial Opinion, Not News Fact
The real signal here isn’t “cheap” — it’s “fast.” Releasing a new generation every three weeks means Google is racing to grab developers through sheer release cadence. The practical takeaway for you and me: don’t hardcode any one model name into your workflow — three weeks from now, it might already be the previous generation.
Watch out hard for “promo pricing.” A lot of people are now calculating “switching over saves half” — but what you’re actually saving is four and a half months’ worth. If what you’re building is meant to run for a full year, budget directly at US$1.50/US$7.50, and treat whatever you save during the promo period as a bonus, not the norm. This isn’t a knock on Google — any time-limited price from any vendor should be calculated this way.
Amo (the one who spots the traps): “The ‘half price’ is written way louder than the part about it doubling back in 2027.” Pimi (the one who finds the upsides): “But the score improvements are real, and the free tier really is accessible — four and a half months is enough to finish a side project.”
Both are right. Test it and use it now, but budget at next year’s price.
VII. Three Things You Can Do Today
- Put “2026-12-31” on your calendar. After that date, your API bill will effectively double — the earlier you know, the earlier you can prepare.
- Move any work that can wait for results over to Batch. The official pricing page explicitly states Batch is half the standard price — this is the only way to cut costs in half without switching models.
- If you use Gemini on mobile, don’t rush to upgrade. The free version today still gives you 3.6 Flash; to use 3.7 Flash in the App you’d need to go through Spark, which means paying for AI Pro (US$19.99/month). Try it for free on AI Studio first before deciding whether to pay.
Sources
Official (this site read each of these directly on 2026-08-14)
- Google official blog, “Gemini 3.7 Flash: our most intelligent workhorse model” (published 2026-08-13): https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/ | Verified 2026-08-14
- Gemini API official pricing page (3.7/3.6 Flash standard price, batch price, free tier): https://ai.google.dev/gemini-api/docs/pricing | Verified 2026-08-14 (read twice on the same day)
- Gemini API official documentation, “What’s new in Gemini 3.7 Flash” (1 million token context, 64K output): https://ai.google.dev/gemini-api/docs/latest-model | Verified 2026-08-14
- Gemini API official documentation, “Tokens” (source of the token-to-character conversion): https://ai.google.dev/gemini-api/docs/tokens | Verified 2026-08-14
- Gemini API official Rate limits page (does not list 3.7 Flash free tier numbers): https://ai.google.dev/gemini-api/docs/rate-limits | Verified 2026-08-14
- Google official blog, “Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber” (published 2026-07-21, source of the 3.6 Flash launch price of US$1.50/US$7.50): https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/ | Verified 2026-08-14
- Gemini subscriptions official website (free tier model, AI Plus/Pro/Ultra list prices): https://gemini.google/subscriptions/ | Verified 2026-08-14
- GitHub official changelog, “Gemini 3.7 Flash is now available in GitHub Copilot” (published 2026-08-13): https://github.blog/changelog/2026-08-13-gemini-3-7-flash-is-now-available-in-github-copilot/ | Verified 2026-08-14
Secondary sources (disclosed per this site’s rule 9.7)
- 9to5Google, “Gemini 3.7 Flash launches three weeks after last model, live in Spark”: https://9to5google.com/2026/08/13/gemini-3-7-flash-launch/ | Verified 2026-08-14
- In this article, the claims that “3.7 Flash on the Gemini App side is only available through Spark (requires AI Pro or Ultra)” and “WebDev Arena Elo 1588 vs 1538” are not sourced from the official website, but cited from the 9to5Google report above; this site is only relaying information — defer to the official announcement.
Cross-verification status (disclosed per rules 9.1/9.2): The US$0.75/US$3.75 promo price, its expiration on 2026-12-31, and its reversion to US$1.50/US$7.50 on 2027-01-01 — these are confirmed across three official pages (Google’s official blog, official API pricing page, and official model documentation), plus one independent outlet, 9to5Google, whose direct reading matches. VentureBeat, Axios, and TechTimes carry consistent headline-level claims, but this site’s attempts to fetch those three pages returned 403 errors, so they could not be read word-for-word and are therefore not listed as cross-verified sources.
Read next
- Gemini Full breakdown: pricing, features and real-world limits
- Gemini Is the free quota enough?
- Gemini Alternatives
- Free plan limits for every AI tool we track
More verified articles on this tool
- [News] Gemini 4 Argon Just Launched — Here's What the Free Tier Pricing Page Actually Says
- Gemini Free Cut 2026-10: Personal Accounts Keep Flash-Lite Only — Pro Gone
- Gemini 4 Argon News 2026: What Free vs Paid Actually Changed at Launch
