Google has dropped its third “Flash” AI model in just six weeks, and the newest one, Gemini 3.8 Flash, is aimed squarely at the developers and teams who want frontier-level coding without paying frontier-level prices. It lands alongside a locked-down cybersecurity sibling, 3.8 Flash Cyber, built for government and enterprise defenders. Here is what changed and who can actually use it.
What Google announced
Google’s official announcement describes Gemini 3.8 Flash as its “most intelligent workhorse model,” a general-purpose reasoning and coding model that follows the 3.6 Flash and 3.7 Flash releases from the last month and a half. The company frames the cadence as deliberate: lighter models shipping fast while it keeps working on the bigger frontier names it teased at I/O, including Gemini 3.5 Pro.
Coding gains that rival the big models
Google says the jump over 3.7 Flash is biggest in software engineering and agentic tasks. Its figures put Gemini 3.8 Flash at 73.7 percent on the DeepSWE v1.1 benchmark for long-horizon coding, just under Claude Opus 5’s 74.0 percent but well ahead of GPT-5.6 Sol (72.7 percent) and the previous 3.7 Flash (65.3 percent).
The performance largely comes from a design choice Google calls working “harder”: on complex tasks the model runs extra reasoning steps and calls tools iteratively, which can mean higher token use at the highest effort levels. That is a detail worth noting for teams watching their API bills, since the per-token price stays low.
- DeepSWE v1.1 (coding): 73.7% vs. Claude Opus 5’s 74.0% and GPT-5.6 Sol’s 72.7%
- HLE-Verified (multi-step reasoning): 54.9%
- Pricing: $0.75 per million input tokens, $3.75 per million output tokens (introductory, same as 3.7 Flash)
Pricing that undercuts the frontier
On price, Gemini 3.8 Flash repeats the introductory rate of 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens. That compares very favorably to Claude Opus 5 at $5 and $25 and GPT-5.6 Sol at $4 and $20 for the same token volumes. The introductory price holds until January 2027, when the regular price rises to $1.50 and $7.50, still a fraction of the competition. Independent tracker Artificial Analysis scores it at 59 on its Intelligence Index, on par with GPT-5.6 Sol and Grok 4.6 at their medium reasoning settings.
Gemini 3.8 Flash Cyber stays locked down
The cybersecurity variant is the bigger security story. Google distributes 3.8 Flash Cyber through its Fairwind Program, a limited-access channel for governments, national cyber authorities, critical-infrastructure operators, and trusted security teams, not for general public use. Google says it beats both the prior 3.5 Flash Cyber and much larger frontier models on autonomous vulnerability discovery, and reports the Chrome Security team found it produced 2.6 times more correct patches to Chrome vulnerabilities than the best far-larger commercial models. It also scores an attack success rate of just 5.5 percent on the Gray Swan prompt-injection benchmark, well under most rivals.
Who can use Gemini 3.8 Flash now
Consumers with a Google AI Pro or Ultra subscription get 3.8 Flash through the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets. Developers can reach it through Google AI Studio and Android Studio, and businesses through Gemini Enterprise.
| Access tier | How to get Gemini 3.8 Flash |
|---|---|
| Consumers (AI Pro / Ultra) | Gemini app, AI Mode in Search, Gemini in Sheets |
| Developers | Google AI Studio, Android Studio, Antigravity |
| Businesses | Gemini Enterprise |
| Cyber variant | Fairwind Program (governments and vetted defenders only) |
For Filipino developers deciding between AI coding assistants, this matters because it is the latest step in Google’s Gemini push after this year’s API billing changes. If you have been comparing it against Anthropic’s Claude, our earlier look at the Claude Fable 5.1 launch and its pricing is a useful counterpart. The takeaway is a familiar one: the smartest model and the most affordable model are no longer the same question.
