Alphabet shares ticked higher in after-hours trading on Tuesday after a report said Google could unveil Gemini 3.8 Flash as soon as Wednesday, September 3, 2026 — and Google confirmed it, officially launching the model on September 2, 2026 at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens. The claim, circulated by outlets including TradingView, BNN Bloomberg, TipRanks and Crypto Briefing, pointed to an internal codename (“Skimaki”), a coding-focused upgrade, and a release cadence that Google itself now describes as its third Flash-tier model in just six weeks. With the launch confirmed, the story has moved from an unverified report to a live, generally available product.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the Report Says About Gemini 3.8 Flash
According to the report — since confirmed by Google — Google’s AI unit prepared and then launched Gemini 3.8 Flash, with the claim traced back to reporting that named the internal codename “Skimaki.” The model is now listed as generally available (GA) under the model code gemini-3.8-flash on Google’s models page, updated September 4, 2026, after moving through production testing on Google’s internal Jetski coding platform, the same environment the company has reportedly used to dogfood earlier Flash releases before they reach the public API. Google’s own Gemini Flash model family page confirms the Flash line’s positioning as the company’s fast, cost-efficient tier, and as of September 2026 it lists Gemini 3.8 Flash as the newest generation in that lineup.
The core strategic angle in the report was coding performance, and Google’s own launch data now backs that up: the company says Gemini 3.8 Flash completes three times as many tasks as Gemini 3.7 Flash in its internal evaluations, a figure Google disclosed alongside the September 2026 release. Google has targeted software-development benchmarks specifically, an area where it has lagged both OpenAI and Anthropic in developer mindshare even as its Gemini 3.7 Flash release closed part of the pricing gap in August. Gemini 3.8 Flash has now shipped under that exact name, with Google’s Interactions API table listing it at “medium” tier and marking it production-ready as of September 3, 2026.
Why GOOGL Stock Gained After-Hours
Alphabet’s Class A shares (GOOGL) reportedly rose to $337.00 in after-hours trading Tuesday, up $1.98, or 0.59%, on the news. The Class C shares (GOOG) moved in step, reportedly reaching $333.62, up $1.59, or 0.48%. Neither move is large by the standards of an earnings reaction, but the fact that both share classes moved together suggests the market was pricing in the strategic implication of the report rather than reacting to a single data point.
Alphabet stock data is tracked continuously on platforms such as TradingView’s GOOGL page, and investors can review Alphabet’s own disclosures through its investor relations site. The after-hours bump is modest, but it lands in a period when investors have been rewarding any signal that Google is closing the perceived coding-model gap with OpenAI and Anthropic, rather than just matching them on price.
Inside the “Skimaki” Codename and Jetski Testing
The report attaches an internal codename, “Skimaki,” to the Gemini 3.8 Flash effort, a detail that first surfaced through leaker channels rather than an official Google blog post. Google has historically used internal codenames for models well before public branding is finalized, and this pattern reportedly held again here.
More notable is the claim that the model was tested on Google’s Jetski developer platform throughout August, and that engineers using it reportedly preferred it to Anthropic’s Claude Opus on coding tasks. That claim originated as a self-reported internal preference relayed through the report, though Google has since shipped the model with a 1-million-token context window and three selectable thinking levels, specs the company detailed in its September 2026 launch documentation. Whether or not the Claude Opus comparison holds up under independent testing, Google is clearly treating coding capability as the primary differentiator for this release rather than general chat performance or multimodal features, which had been the focus of recent Flash-tier updates.
Google’s Accelerating Flash Release Cadence in 2026
Gemini 3.8 Flash’s launch capped off an unusually fast release schedule. Google shipped Gemini 3.6 Flash on July 21, 2026, then followed with Gemini 3.7 Flash on August 13, 2026, a gap of just 23 days, before confirming Gemini 3.8 Flash on September 2, 2026 — a release Google itself describes as its third Flash model in just six weeks.
| Model | Reported Release Date | Gap From Prior Version | Status |
|---|---|---|---|
| Gemini 3.6 Flash | July 21, 2026 | Baseline | Officially released |
| Gemini 3.7 Flash | August 13, 2026 | 23 days | Officially released, marketed as Google’s “most capable Flash model” |
| Gemini 3.8 Flash | Reportedly as soon as September 3, 2026 | ~21 days (if confirmed) | Unconfirmed by Google; based on report only |
A monthly-or-faster release rhythm for a flagship-adjacent product tier is unusual even by 2026 AI-industry standards, where major labs typically space large model updates months apart. Google’s own AI technology blog has chronicled the Gemini program’s rapid iteration pace this year, though it has not yet posted anything referencing a 3.8 release.
Gemini 3.6 Flash and 3.7 Flash: What Came Before
Gemini 3.7 Flash, released just three weeks before this report, was positioned by Google as the company’s most capable Flash-tier model to date, with an emphasis on coding-agent workflows. Coverage at the time, including reporting picked up by TradingView, framed 3.7 Flash as aimed squarely at developers building autonomous coding agents rather than general consumer chat use. Google’s Vertex AI documentation for Gemini 3.7 Flash lists the model’s availability through the Vertex AI platform for enterprise customers.
That release followed Gemini 3.6 Flash by just over three weeks, itself part of a broader July rollout that included multiple new Gemini variants. The pattern across both releases: incremental capability gains shipped fast, rather than a single large jump shipped rarely. If the Gemini 3.8 Flash report is accurate, it would represent a continuation, not a break, from that strategy.
Pricing: What Developers Are Paying Right Now
Google’s official pricing for Gemini 3.6 Flash, 3.7 Flash, and now the newly launched Gemini 3.8 Flash sits at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens, in effect through December 31, 2026, according to Google Cloud’s published pricing pages. Google has confirmed that rate will double to $1.50 per million input tokens and $7.50 per million output tokens starting January 1, 2027 — the same step-up schedule the company laid out when Gemini 3.8 Flash launched on September 2, 2026.
| Model | Input Price (per 1M tokens) | Output Price (per 1M tokens) | Pricing Window |
|---|---|---|---|
| Gemini 3.6 Flash | $0.75 | $3.75 | Introductory, through Dec. 31, 2026 |
| Gemini 3.7 Flash | $0.75 | $3.75 | Introductory, through Dec. 31, 2026 |
| Gemini 3.6/3.7 Flash (post-introductory) | $1.50 | $7.50 | Effective Jan. 1, 2027 |
| Gemini 3.8 Flash (reported) | Not yet disclosed | Not yet disclosed | Pending official launch |
Pricing for Gemini 3.8 Flash is no longer speculative: Google has confirmed the same $0.75/$3.75 introductory tier used for its Flash predecessors, and Google Cloud additionally lists a cached-input rate of just $0.075 per million tokens, globally available through December 31, 2026. That matches the pattern the company had already used for 3.6 and 3.7 Flash, rather than undercutting it further.
The Coding Performance Gap Google Is Trying to Close
The report frames Gemini 3.8 Flash’s core mission as closing a coding-capability gap against OpenAI and Anthropic. That framing matters because coding has become the primary battleground for frontier and near-frontier models in 2026: enterprise buyers increasingly select a model provider based on how well it performs inside agentic coding tools, not general chatbot quality.
Google has invested heavily in developer-facing tooling this year, and products like Gemini API access for developers have expanded alongside the Flash release cycle. A coding-focused Gemini 3.8 Flash, if it materializes as described, would extend that push rather than start a new one. Rival labs have made similar coding-first arguments for their own releases, including OpenAI’s GPT-5.6 family and Anthropic’s Claude Opus line, both of which have been marketed heavily around agentic and coding use cases in 2026.
How Gemini 3.8 Flash Reportedly Compares to Claude Opus
The most eye-catching detail in the report is the claim that Google’s internal testers preferred Gemini 3.8 Flash over Anthropic’s Claude Opus on coding tasks during Jetski testing. That claim comes from the same report and has not been independently verified through a public benchmark suite. Anthropic’s own Claude product page continues to market the Opus tier as its top-end coding and reasoning model, and the company has not commented on the comparison described in the report.
It is worth noting that internal preference testing, run by the company that built the model being tested, is a different evidentiary bar than a neutral, reproducible benchmark. Readers should treat the Claude Opus comparison as a company-sourced claim relayed through a report, not as an independently confirmed result, until Google publishes model card data or a third party runs a head-to-head evaluation.
Competitive Landscape: OpenAI, Anthropic and the Flash Wars
Google’s rapid Flash cadence sits inside a broader 2026 pattern where all three major U.S. labs have compressed their release cycles. OpenAI has continued iterating on its GPT-5.6 model family, while Anthropic has faced its own reported model developments, including leaked details about upcoming Claude models tied to its reported IPO ambitions. DeepSeek and other open-weight labs out of China have added further pricing pressure, with comparisons like the one covered in our DeepSeek V4-Flash vs. Gemini 3.7 Flash vs. Qwen3.8-Flash-Next breakdown showing how far token pricing has fallen across the industry this year.
What makes the Gemini 3.8 Flash report notable in that context is speed, not novelty. Google is not claiming a categorical leap in capability; it is reportedly claiming a fast, iterative improvement cycle that keeps it visible in developer conversations every three to four weeks. That is a different competitive strategy than a large, infrequent flagship launch, and it changes how enterprise buyers evaluate vendor lock-in and model migration costs.
Market Reaction and Analyst Context
The after-hours move in GOOGL and GOOG was modest in percentage terms, well under a full point, which is consistent with a market treating this as an incremental positive rather than a game-changing catalyst. Multiple outlets picked up the story within the same news cycle, a distribution pattern that itself tends to amplify short-term price sensitivity even when the underlying news is unconfirmed by the company involved.
| Outlet | Core Claim Reported | Sourcing Note |
|---|---|---|
| TradingView | Google preparing Gemini 3.8 Flash launch, GOOGL/GOOG after-hours gains | Cited original reporting on the release timeline |
| BNN Bloomberg | Syndicated coverage of the Gemini 3.8 Flash report | Part of broader financial-news distribution of the story |
| TipRanks | Market-news coverage tying the report to Alphabet share movement | Financial-market focused coverage |
| Crypto Briefing | Detailed the “Skimaki” codename, Jetski testing, and pricing cadence | Published September 1, 2026 |
None of the coverage cited here includes an on-record statement from Google confirming the model name, launch date, or specifications. That gap is the central caveat around this entire story: it is a report about an anticipated release, not a company announcement.
Historical Context: Google’s AI Model Cadence Since 2024
Google’s Gemini program has steadily compressed its release intervals since its original 2023 debut. Early Gemini generations shipped on multi-month cycles, with the company using major developer conferences as launch moments. By 2026, that pattern has shifted toward continuous, incremental Flash-tier releases punctuated by less frequent flagship-tier updates, a strategy that mirrors how mobile operating systems and cloud services ship point releases rather than annual rewrites.
This shift reflects a broader industry trend: as training and fine-tuning pipelines mature, the marginal cost of shipping a Flash-tier update has fallen relative to the marginal cost of shipping a flagship model. That economic reality, more than any single competitive threat, likely explains why Google can plausibly ship a third Flash-tier model within seven weeks without straining its infrastructure or research teams beyond what a normal quarter would require.
What This Means for Developers and Enterprises
For developers currently building on Gemini 3.7 Flash, a Gemini 3.8 Flash release would likely mean another migration decision within a matter of weeks, evaluating whether the newer model’s coding improvements justify updating production pipelines, prompts, and evaluation suites. Given the introductory pricing structure Google has used for 3.6 and 3.7 Flash, it is reasonable to expect a similarly aggressive price point if 3.8 Flash follows the same commercial playbook, though that remains unconfirmed.
Enterprise teams evaluating multi-model strategies should treat this report as a signal to watch rather than an action item. Committing infrastructure changes based on an unconfirmed model name and launch date carries obvious risk if the actual release slips, is renamed, or ships with different capabilities than reported.
Risks: Why “As Soon As Wednesday” Isn’t Confirmed
The phrase “as soon as Wednesday” is itself a hedge, not a commitment. Reports describing internal codenames and testing platforms ahead of an official announcement have, in past AI-industry cycles, sometimes preceded delayed launches, renamed products, or scaled-back feature sets. Nothing in the current reporting guarantees Google will use the name “Gemini 3.8 Flash,” ship on September 3, or match the coding-performance claims attributed to internal testing.
Readers should also note that the stock reaction described here happened before any official launch, meaning it reflects market expectations about a rumor, not a confirmed product event. If Google does not ship on the reported timeline, any further stock movement tied to a delay or a smaller-than-expected update would be a separate, distinct data point.
Predictions: What Happens Next
- Google is likely to confirm or deny the Gemini 3.8 Flash report within days, given how quickly the story has spread across financial and tech outlets.
- If the model ships, expect Google to lean heavily on coding-agent benchmarks in its announcement, continuing the framing used for Gemini 3.7 Flash in August.
- Pricing, if disclosed, will most likely match or undercut the current $0.75/$3.75 introductory tier to keep pressure on OpenAI and Anthropic on cost-per-token.
- Expect Anthropic and OpenAI to respond with their own coding-focused updates or marketing pushes within the following weeks, continuing the compressed release-cycle pattern seen throughout 2026.
- GOOGL’s after-hours reaction is unlikely to translate into a sustained rally on its own; watch for confirmation of actual coding-benchmark gains as the more durable catalyst for the stock.
The Bottom Line for Google’s AI Strategy
With Gemini 3.8 Flash officially confirmed and shipped under that exact name, the underlying signal is consistent with everything Google has done since July: ship fast, price aggressively, and target coding workloads specifically rather than general-purpose chat improvements. That strategy has now produced three confirmed releases, Gemini 3.6 Flash, Gemini 3.7 Flash, and Gemini 3.8 Flash, inside a six-week window. Rather than a surprising escalation, it is a continuation of a pattern the company had already established and that the market had already started pricing in before the September 2, 2026 launch.
Frequently Asked Questions
Is Gemini 3.8 Flash officially confirmed by Google?
Yes. Google officially launched Gemini 3.8 Flash on September 2, 2026, and lists it as generally available (GA) under the model code gemini-3.8-flash on its models page, last updated September 4, 2026. The initial reporting from outlets including TradingView, BNN Bloomberg, TipRanks and Crypto Briefing turned out to match the actual release.
When could Gemini 3.8 Flash launch?
The report said Google could unveil the model “as soon as Wednesday,” referring to September 3, 2026, and Google beat that timeline slightly, officially launching Gemini 3.8 Flash on September 2, 2026.
Why did GOOGL stock rise on this news?
Alphabet’s GOOGL shares reportedly rose to $337.00 after-hours, up 0.59%, while GOOG shares reportedly rose to $333.62, up 0.48%. The move reflects investor interest in Google’s accelerating AI release cadence rather than any confirmed earnings or product event.
What is the “Skimaki” codename?
“Skimaki” is the internal codename reportedly tied to the Gemini 3.8 Flash development effort, according to the report. Google has not publicly confirmed this codename.
How does Gemini 3.8 Flash compare to Claude Opus?
The report claims Google’s internal testers on the Jetski platform preferred Gemini 3.8 Flash over Anthropic’s Claude Opus for coding tasks. This is a company-sourced internal claim relayed through the report, not an independently verified benchmark result.
How much does Gemini 3.7 Flash cost right now?
Gemini 3.7 Flash is priced at an introductory rate of $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026, after which Google’s published pricing steps up to $1.50 and $7.50 per million tokens, respectively, starting January 1, 2027.
How fast has Google been releasing new Gemini Flash models in 2026?
Google released Gemini 3.6 Flash on July 21, 2026, and Gemini 3.7 Flash on August 13, 2026, a gap of 23 days. A Gemini 3.8 Flash release on the reported September 3 date would put roughly three weeks between the two most recent versions, continuing the same pace.
Should developers switch to Gemini 3.8 Flash immediately if it launches?
Not necessarily. Enterprises should evaluate any new model release against their own benchmarks and cost requirements before migrating production workloads, especially since key details such as final pricing and independently verified performance data were not available at the time of the report.
