Gemini 3.8 Flash has a price with an expiry date. Until December 31, 2026, Google charges an introductory $0.75 per million input tokens and $3.75 per million output tokens. From January 1, 2027, that doubles to $1.50 and $7.50. If you're building on it now, budget for the second number.
Google announced the model on September 2, 2026, and it's generally available in the Gemini API as gemini-3.8-flash.
Pricing, Specs and Token Usage
| Item | Value |
|---|---|
| Introductory price (until Dec 31, 2026) | $0.75 in / $3.75 out per 1M tokens |
| Standard price (from Jan 1, 2027) | $1.50 in / $7.50 out per 1M tokens |
| Context | 1,048,576 input / 65,536 output tokens |
| Modalities | Text, image, video, audio, PDF in; text out |
| Thinking | low, medium (default), high; minimal errors |
| Serving | Batch API, Flex, Priority |
| Knowledge cutoff | March 2026 (model card) |
Sources: the developer guide, the model page and the pricing page. Google also warns that 3.8 Flash "can use more tokens on longer running and complex tasks, by design," and suggests a lower thinking level for everyday work. That's a cost you won't see in the per-token rate.
Official Claims and Arena Votes
Google's announcement calls it its "most intelligent workhorse model", better than Gemini 3.7 Flash at software engineering, agentic tasks and multi-step reasoning. The DeepMind page reports 54.9% on HLE-Verified, a Google figure.
Arena's blind votes mostly agree. On the October 8, 2026 text snapshot (style control), gemini-3.8-flash-high ranked #8 of 414 at 1497 (1492–1501) on 31,437 votes, ten places above Gemini 3.7 Flash (high) at #18 and 1486.
| Category | Rank | Rating (interval) | Votes |
|---|---|---|---|
| Math | #4 | 1522 (1507–1538) | 1,584 |
| Creative writing | #7 | 1487 (1479–1495) | 6,851 |
| Multi-turn | #8 | 1503 (1494–1512) | 4,708 |
| Chinese | #12 | 1539 (1526–1551) | 2,513 |
| Instruction following | #14 | 1488 (1482–1495) | 11,246 |
| Coding | #21 | 1530 (1523–1538) | 8,355 |
The coding rank is the surprise, given how much of Google's pitch is about software engineering. Only the high thinking setting is on the board. Data comes from the Arena text leaderboard (CC BY 4.0) via AI Rank.
→ Track it on AI Rank: Gemini 3.8 Flash (high) — Arena rating, rank and 30-day history
OpenRouter Usage
google/gemini-3.8-flash-20260902 was #14 on October 11 with 199B tokens (−6%), #15 for the week of October 5–11 with 1.71T (−5%), and #11 for the month to October 11 with 7.48T (+292%). The separate …:batch ID was #215 for the week at 2.78B. Direct Google API traffic isn't counted.
→ Live usage: See where it ranks this week in the OpenRouter model usage ranking on AI Rank (tokens, Oct 5–11, 2026 window).
Availability
Developers get it in Google AI Studio, Android Studio and Google Antigravity, where it is also the default model for the Antigravity agent in Gemini Managed Agents. Enterprises use Gemini Enterprise. Google AI Pro and Ultra subscribers see it in the Gemini app, AI Mode in Search and Gemini in Sheets. A separate Gemini 3.8 Flash Cyber went only to trusted defenders through Google's Fairwind Program.
Compare before you choose: pricing changes and rankings move daily — check the current model rankings on AI Rank before you commit.
Or 📱 get the AI Rank app to follow them on your phone.
Prepared with automated editorial assistance and checked against official Google sources and dated AI Rank data (Arena snapshot Oct 8, 2026; OpenRouter windows ending Oct 11, 2026) on October 12, 2026. We did not run our own benchmarks for this article.