Gemini 3.8 Flash's introductory price doubles on December 31, 2026
Google's third Flash release in six weeks keeps the $0.75/$3.75 rate card. But the model also spends more tokens per task. Both numbers in your cost model are moving before the price even changes.
The Engineer · Build desk
What happened
- Google launched Gemini 3.8 Flash and 3.8 Flash Cyber on Wednesday, its third Flash release in six weeks, with 3.7 Flash only three weeks old at the time.
- Gemini 3.8 Flash keeps 3.7's introductory price of $0.75 per million input tokens and $3.75 per million output tokens, which expires on December 31, 2026 and rises to $1.50/$7.50.
- On an internal Google benchmark covering codebases in 20 programming languages, 3.8 Flash Cyber exceeds a 70% vulnerability discovery success rate.
Why it matters
- What it costs cost Two variables move at once for anyone forecasting spend: the published rate doubles at a known date, and tokens per task rise by an unpublished amount, so a budget carried over from 3.7 understates the bill twice.
- Who has to decide decision Google's own guidance makes staying on 3.7 Flash a supported cost decision rather than deferred migration work, which changes how a version pin is defended at review.
- Who's exposed exposure Frontier patching capability is now reachable only by buying from a Fairwind partner or being one, so everyone else meets it as a feature in someone else's product.
- What doesn't add up contradiction Google's comparison set stops at OpenAI and Anthropic's cheaper tiers, and the numbers The New Stack adds for Fable 5.1 and the Chinese models point at a different price/performance conclusion.
Next story · Invest · Oct 2
Google prices its third-ranked Gemini 4 Argon at half the cost of Anthropic's Claude Opus 5.5Confirmed12 publishers
Billing is per token, not per task. That is what makes the unchanged rate card misleading. Google's own copy says 3.8 Flash executes extra reasoning steps and calls tools iteratively, and may use more tokens to maximise performance, especially at higher effort levels [3]. Same rate, more units per unit of work.
The config line that proves Google knows this sits in the same post: use lower effort levels to minimise token overhead, or stay on 3.7 Flash, which remains fully supported for efficiency-first workloads [4]. A vendor pointing you at last month's model for cost reasons is doing you a favour. It is also telling you what the new one costs.
Then the date. Both halves of the introductory price double: input from $0.75 to $1.50, output from $3.75 to $7.50 per million tokens, exactly 2.0x each [2][18]. A unit of work that reads a million input tokens and writes a million output tokens costs $4.50 now and $9.00 after the expiry [19]. If the extra diligence lifts token use by half, the post-expiry bill is 3x what you are paying today [20]. That 50% is my illustration, not Google's number; nothing in the announcement quantifies the overhead. Google also says the release cadence is accelerated by long-running agentic loops that recursively evaluate and refine the models [9], which is a reason to expect another version before you finish migrating to this one.
On benchmarks, treat the table as a claim about someone else's repo. Google says 3.8 Flash matches Opus 5 on DeepSWE and beats GPT-5.6 Sol and Sonnet 5 [23]. For that to transfer, your tasks need to look like DeepSWE's: long-horizon engineering problems solved end to end with a verifiable finish. Where it still trails is computer use, 59% against 75.4% for Opus 5 on OSWorld-2.0 [14], and knowledge work, 1545 on GDPVal against 1824 for Opus 5, 279 points back [15][22]. The New Stack notes that Google's comparison set omits Anthropic's Fable 5.1, which shipped a day earlier [11]; on Terminal-Bench 4.0, Flash 3.8 scores 19.1% against Fable's 55.8% and Opus 5's 51.8% [10], a 2.9x gap to Fable [21]. On the coding-focused Terminal-Bench 2.1, Flash 3.8 beats its competitors [12], and The New Stack reads Terminal-Bench 4.0 as the outlier rather than the verdict [13]. The same publication points out that GLM-5.3, DeepSeek v4 Pro and Kimi K3 sit in the same league on DeepSWE 1.1 at better price/performance [24], which is the comparison the price change actually invites.
Fairwind is the more interesting piece of engineering discipline. Google says it invested in vulnerability fixing from the start and prioritised it over offensive capabilities like exploitation [17], and reports 47.2% pass@1 on Collinear's CWE-Bench against 47.8% for a leading frontier model at significantly lower cost [6]. Near-parity on patching at Flash cost, still under 50% first-try, means a human reads every diff. Access runs through about 650 trusted partners including CrowdStrike, Palo Alto Networks, Datadog, Snowflake, Wiz, Accenture and the Center for Internet Security [5]. The 3.5 Cyber equivalent was unnamed and, by The New Stack's reading, ad hoc [16]. Naming the programme and putting 650 partners behind it turns a capability into a procurement category.
What to watch
- Whether Google extends the introductory price past December 31, 2026 or lets both rates double as published.
- Any published token-per-task comparison between 3.7 and 3.8 Flash at matched effort levels, which is the figure every cost model is currently guessing.
- Whether Fairwind admission criteria are published, or the 650-partner count stays a round number in a blog post.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence58
- Adoption35
- Hype gap+22
- Incentives72
- Confidence62
Perspective Coverage
8 publishers- Builder
- Builder 53%
- Operator
- Operator 29%
- Investor
- Investor 18%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Google launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber on Wednesday; it is the company's third Flash release in six weeks, and Gemini 3.7 Flash is only three weeks old.
- [2]
Gemini 3.8 Flash is available at $0.75 per million input tokens and $3.75 per million output tokens, the same introductory price as 3.7 Flash; the introductory price expires on December 31, 2026 and then goes to $1.50/$7.50.
- [3]
Google says 3.8 Flash "works harder": on complex tasks it exhibits greater diligence, executing extra reasoning steps and calling tools iteratively, and at times may use more tokens to maximise performance, especially at higher effort levels.
- [4]
Google tells developers that where compute efficiency is the primary constraint they can use lower effort levels to minimise token overhead, or continue to rely on Gemini 3.7 Flash, which remains fully supported for efficiency-first workloads.
- [5]
Gemini 3.8 Flash Cyber is available only to trusted defenders through Google's new Fairwind Program, comprising about 650 trusted partners including Accenture, CrowdStrike, the Center for Internet Security, Datadog, Palo Alto Networks, Snowflake and Wiz.
- [6]
On CWE-Bench, an external patching benchmark run by Collinear, Gemini 3.8 Flash Cyber records a pass@1 of 47.2% compared with 47.8% for a leading frontier model, at significantly lower cost.
- [7]
Four Flynn, Google's VP for Security and Privacy, writes that the defender's edge comes from shrinking the time between detecting a flaw and patching it, and that Fairwind gives government and enterprise partners autonomous tools to repair systems faster and at scale.
- [8]
On an internal Google benchmark requiring discovery of vulnerabilities across complex codebases spanning 20 programming languages, 3.8 Flash Cyber reaches a success rate exceeding 70%.
- [9]
Google says both releases are powered by the same foundational intelligence and further accelerated by long-running agentic loops designed to recursively evaluate and refine the underlying models.
- [10]
On Terminal-Bench 4.0, Flash 3.8 scores 19.1%, Anthropic's Fable 5.1 scores 55.8% and Opus 5 scores 51.8%.
- [11]
Google's published benchmarks do not include Anthropic's Fable 5.1, which was released a day earlier and which The New Stack describes as a much stronger but far more expensive model.
- [12]
On Terminal-Bench 2.1, which focuses on coding, Flash 3.8 does better than its competitors.
- [13]
The New Stack describes Terminal-Bench as the outlier in the comparison, noting the Gemini model does quite well on other agentic tasks even against flagship models.
- [14]
On computer use, Flash 3.8 scores 59% on OSWorld-2.0 against 75.4% for Opus 5.
- [15]
On GDPVal, a benchmark of knowledge work tasks, Google scores 1545 against Opus 5 at 1824 and Sonnet 5 at 1584.
- [16]
With Flash 3.5 Cyber, Google ran a limited-access programme that had no name and, according to The New Stack, seemed a bit ad hoc at the time.
- [17]
Google says it invested in vulnerability fixing from the start with 3.8 Flash Cyber and prioritised it over offensive capabilities like exploitation.
- [18]
The post-expiry prices are exactly 2.0x the introductory prices on both input and output.
- [19]
A unit of work consuming one million input tokens and one million output tokens costs $4.50 at the introductory price and $9.00 after December 31, 2026.
- [20]
If token consumption per task rose by 50%, the post-expiry cost per task would be 3x the current cost per task. The 50% figure is an illustration, not a sourced number.
- [21]
Fable 5.1's Terminal-Bench 4.0 score is about 2.9 times Flash 3.8's.
- [22]
Google's GDPVal score is 279 points below Opus 5's.
- [23]
Google says 3.8 Flash often outperforms GPT-5.6 Sol, Claude Sonnet 5 and Opus 5 on complex engineering tasks, and on DeepSWE it matches Opus 5 while beating GPT-5.6 Sol and Sonnet 5.
- [24]
The New Stack notes that on benchmarks like DeepSWE 1.1, Chinese models including GLM-5.3, GLM-5.3 Flash, DeepSeek v4 Pro and Kimi K3 are in the same league and often offer better price/performance ratios.
Sources
8 independent publishers whose own reporting we read for this story.
- archive.thedeepview.comGemini 3.8's edge is intelligence per dollar
1 article · September 3, 2026
- blog.googleIntroducing Gemini 3.8 Flash and 3.8 Flash Cyber
2 articles · September 2, 2026
- blog.vercel.comGemini 3.8 Flash now available on AI Gateway
1 article · September 1, 2026
- dev.toGoogle Locked Its AI Vulnerability Hunter Behind 650 Seats. Build Your Own for Under a Cent per File
1 article · September 4, 2026
- simonwillison.netRelease: llm-gemini 0.34
1 article · September 2, 2026
- testingcatalog.comGoogle releases Gemini 3.8 Flash and Flash Cyber
2 articles · September 3, 2026
- the-decoder.comGemini 3.8 Flash is Google's third budget model in six weeks while frontier models remain MIA
2 articles · September 2, 2026
- thenewstack.ioGoogle ships its third Gemini Flash model in six weeks
2 articles · September 2, 2026