Gemini 3.8 Flash
price and who can get it
Google shipped Gemini 3.8 Flash and 3.8 Flash Cyber on 2026-09-02. The published entry price is $0.75 per 1M input tokens and $3.75 per 1M output, and the post states plainly that this is an introductory price expiring 2026-12-31, after which it becomes $1.50 / $7.50. The Cyber variant is not open to general developers — it goes only to trusted defenders through the Fairwind Program. This page is about which tier you can actually get.
Updated 2026-09-09
Four official numbers
Input / output per 1M tokens
The official blog calls this an introductory price, not the standing rate.
Last day of the intro price
From the next day the official rate becomes $1.50 / $7.50. Any budget model for this model has to include that date.
The 2027 increase
Input $0.75→$1.50 and output $3.75→$7.50 — both exactly double. That is published, not inferred.
The only route to the Cyber variant
3.8 Flash Cyber is limited to trusted defenders through the Fairwind Program; general developers cannot get it.
Two models, two very different availabilities
This launch is a pair. Gemini 3.8 Flash is for general developers, reachable through Google AI Studio, Android Studio and Google Antigravity, with enterprises going through Gemini Enterprise and consumers through Google AI Pro and Ultra subscriptions. 3.8 Flash Cyber is a restricted security-oriented build that the announcement limits to trusted defenders via the Fairwind Program. When you see capability claims about "3.8 Flash", check which one is meant — their availability is not the same.
The line in the pricing that gets dropped
Alongside $0.75 / $3.75 the announcement states that the "introductory price expires on December 31, 2026" and gives the post-2027 rates directly: $1.50 / $7.50. If you are building a cost model that crosses the year boundary, using today's price understates it by half. Most second-hand write-ups only copy the first half of that sentence.
How the Flash line has moved
Google ships a batch of Gemini 3.6 Flash models. The same official post states that Gemini 3.5 Pro is still testing with partners and will be made broadly available as soon as it is ready, and announces that the most ambitious pre-training run yet has begun, for Gemini 4.
Gemini 3.8 Flash and 3.8 Flash Cyber ship. Entry price $0.75 / $3.75 with the introductory rate expiring 2026-12-31. The Cyber variant is limited to the Fairwind Program.
This page was checked against the official blog. As of that date the announcement gives no context-window figure for 3.8 Flash and does not update the status of Gemini 3.5 Pro or Gemini 4.
What the announcement says and does not say
✅ In the official announcement
Release date 2026-09-02; two models, Gemini 3.8 Flash and 3.8 Flash Cyber; pricing $0.75 / $3.75 per 1M tokens with the introductory rate expiring 2026-12-31 and $1.50 / $7.50 after; distribution through Google AI Studio, Android Studio, Google Antigravity, Gemini Enterprise and Google AI Pro / Ultra; and the Cyber variant limited to trusted defenders through the Fairwind Program.
⚠️ What this announcement does not give
The context-window size, an explicit API model id string, and the current status of Gemini 3.5 Pro and Gemini 4 are all absent from this post. Specific ids and context figures circulating online do not appear in it and are therefore unverified against this source; check Google's official model documentation before relying on them. We have not put those numbers on this page.
Worth switching from the previous generation?
3.8 Flash
The newest generation at an official entry price of $0.75 / $3.75 — but remember that is the rate through year end. Suits teams running volume now whose budget model can absorb the 2027 increase.
Staying on 3.7 / 3.6 Flash
If your workflow is already tuned on an older id, benchmark on the same real requests before moving. The price advantage halves after 2026-12-31.
How to decide whether to migrate
Three steps. First, put the post-2027-01-01 rate into your annual model rather than estimating on $0.75 / $3.75. Second, confirm you mean 3.8 Flash and not the Cyber variant, which general developers cannot obtain. Third, benchmark on your own real requests — differences within the Flash line show up per task, and a leaderboard position does not convert into your invoice.
On QCode
To be explicit: the Gemini family currently has no real call volume on QCode, so we make no availability claim for it. What QCode actually carries are the Claude, GPT and Chinese open-weight lines — models with real traffic over the last 30 days include claude-sonnet-5, gpt-5.6-sol, gpt-6-astra, deepseek-v4-flash, glm-5.3-flash, kimi-k3 and qwen3.8-flash. If what you are evaluating is running several vendors behind one key, those are the ones that are actually callable today.
Frequently asked questions
What does Gemini 3.8 Flash cost?
The official announcement gives $0.75 per 1M input tokens and $3.75 per 1M output, and states explicitly that this is an introductory price expiring 2026-12-31. From 2027-01-01 it becomes $1.50 / $7.50 — both figures double.
Can I use 3.8 Flash Cyber?
Most likely not. The announcement limits it to trusted defenders through the Fairwind Program; it is not a tier a general developer can self-serve.
What is the context window?
The official announcement does not give a figure. Numbers circulating online do not appear in it, so this page does not adopt them. Check Google's official model documentation for the exact value.
What is the status of Gemini 3.5 Pro and Gemini 4?
This 3.8 Flash post does not mention either. What is checkable is the 2026-07-21 official blog: 3.5 Pro was still testing with partners and would be made broadly available as soon as it was ready, and Google announced it had begun its most ambitious pre-training run yet, for Gemini 4. The 09-02 post does not update those.
Can I call Gemini on QCode?
Not as an availability claim today. The Gemini family has no real call volume in our 30-day usage statistics, so this page makes no availability promise. What QCode actually carries right now are the Claude, GPT and Chinese open-weight lines.
So what is this page for?
Two things. It pins down the expiry of the introductory price and the actual 2027 rate so a cross-year budget is not understated by half, and it separates 3.8 Flash from the Cyber variant that only goes to trusted defenders, so you do not choose based on capabilities you cannot access.
Sources
The official Google launch blog for Gemini 3.8 Flash and 3.8 Flash Cyber (2026-09-02), and the 2026-07-21 official blog covering Gemini 3.6 Flash and the status of 3.5 Pro and Gemini 4. Captured 2026-09-09. Where the official announcement gives no figure, this page marks it as unpublished rather than filling in a third-party number. QCode call volumes come from this site's own 30-day usage statistics.
One key for the vendors that are actually callable
Claude, GPT-6 Astra, the three GPT-5.6 tiers and the Chinese open-weight lines share one QCode key and one quota; switching is a model-id change. Plans start at ¥60/month and the key is live the moment you pay.
Related reading
Gemini 3.7 Flash guide
Positioning, price and fit for the previous Flash generation.
Gemini 3.5 Pro tracker
Why the flagship line still has no GA, and what Google has actually said.
Gemini 3.6 Flash guide
What that generation shipped with and the problem it addressed.
Every figure on this page comes from Google's official launch blog, captured 2026-09-09; Google may change them, so defer to the official page. The Gemini family has no real call volume on QCode and this page is not an availability claim.