On September 2, 2026, Google introduced Gemini 3.8 Flash and 3.8 Flash Cyber—the third Flash release in six weeks, by Google’s count, after 3.7 Flash three weeks earlier. Authors on the note are Tulsee Doshi and Raluca Ada Popa. 3.8 Flash is listed at the same introductory price as 3.7 Flash: $0.75 per million input tokens and $3.75 per million output tokens. Those figures are USD as published. CAD is on the invoice. This desk is not inventing a Canadian list price.
What Google listed for Gemini 3.8 Flash
The Google blog is the primary note for price and product framing. Google calls 3.8 Flash its most intelligent workhorse model, with gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning in specialized domains. It says the model can use more tokens at higher effort; for efficiency-first work, use lower effort levels or keep 3.7 Flash, which remains fully supported.
“It is available at the same introductory price as 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens.”Google blog, Introducing Gemini 3.8 Flash and 3.8 Flash Cyber, September 2, 2026
Footnote 1 on that page: introductory price expires on December 31, 2026. Starting January 1, 2027, $1.50 per million input tokens and $7.50 per million output tokens will apply. Quote the footnote when you quote the intro rate.
The optional Gemini 3.8 Flash model card is dated published 2 September 2026. It notes a knowledge cutoff of March 2026, with some domains limited to January 2025 in line with the Gemini 3 Model Family. Google says assessment under its Frontier Safety Framework (April-2026) found the model did not reach any Tracked or Critical Capability Levels (T/CCLs).
What it costs
These are Google’s published USD rates from the Sep 2 blog. Do not quote them as CAD. CAD is on the invoice.
| Model | Input / 1M | Output / 1M | Notes |
|---|---|---|---|
| Gemini 3.8 Flash (intro) | $0.75 | $3.75 | Same intro as 3.7 Flash through Dec 31, 2026. |
| Gemini 3.8 Flash (from Jan 1, 2027) | $1.50 | $7.50 | Google footnote after intro expires. |
| GPT-6 Astra (compare) | $10 | $50 | OpenAI flagship. Not this release. |
| Claude Fable 5.1 (compare) | $10 | $50 | Anthropic flagship. Not this release. |
Developers: Google says build with 3.8 Flash in Google Antigravity, Gemini API via Google AI Studio and Android Studio, or Stitch. Enterprises: 3.8 Flash in Gemini Enterprise. Consumers: Google AI Pro and Ultra across the Gemini app, AI Mode in Search, and Gemini in Sheets. That is Google’s availability list, not an Alberta inventory.
Flash Cyber is Fairwind only
Gemini 3.8 Flash Cyber is Google’s cybersecurity model for vulnerability detection and automated patching. The blog says it is available to trusted defenders through the new Fairwind Program—trusted government authorities, critical infrastructure operators, and software maintainers with prioritized access. It is not a generally available Alberta shop SKU. Do not order it as the default automation path.
On CyberGym, Google says Flash Cyber shows frontier-level autonomous vulnerability discovery. On an internal multi-language vulnerability-discovery benchmark spanning 20 programming languages, Google claims a success rate exceeding 70%. On CWE-Bench (Collinear), Google’s blog wording: pass@1 of 47.2% compared to a leading frontier model at 47.8%, at significantly lower cost. This desk prefers that Google blog wording. DeepMind’s Gemini Cyber page and Fairwind Program page name Fable 5 at 47.8% on the same CWE-Bench comparison—cite that as a second source, not a rewrite of the Google blog lead sentence.
Fairwind FAQ (on the Fairwind page): when accessed directly as a managed model on Gemini Enterprise Agent Platform, Gemini 3.8 Flash Cyber supports zero data retention. Until that ZDR enterprise path is confirmed for the shop, treat the default as a US/cloud data path.
What Alberta operators should do
Alberta work here means AI automation Alberta and private AI security. Flash at $0.75 / $3.75 is the shop SKU against Astra or Fable at $10 / $50 when the job fits a workhorse, not a flagship. Pick on task fit and data path, not a fake price war across vendors.
Flash Cyber is Fairwind / trusted defenders—not every Edmonton shop. Do not point cyber models at client networks, tenders, or credentials without authorization. Keep that traffic on a private AI security path. If the workflow has to stay on the desk, that is AI consulting work, not a public paste.
US/cloud is the default data path unless a ZDR enterprise path is confirmed. A Flash price card does not put inference in Alberta.
The guardrail
Price is not permission. A $0.75 / $3.75 card does not make Flash Cyber generally available, and it does not authorize scanning anyone else’s network. For AI automation Alberta work, use 3.8 Flash when the job fits, keep 3.7 Flash or lower effort when efficiency is the constraint, and leave Fairwind cyber capability to trusted defenders who have the paperwork.