What is Gemini 3? 4 AI plans, latest models and free use

Gemini 3 is Google's family of AI models, launched on November 18, 2025. As of October 2026, the newest Flash model is Gemini 3.8 Flash and the newest Pro model in the API is Gemini 3.1 Pro Preview. You can still use the Gemini app without paying, but with fewer model choices.

Gemini 3 is a model family Google started on Nov 18, 2025

On November 18, 2025, Google announced Gemini 3 in a blog post, calling it its most intelligent model, with a 1 million token context window. The first release was Gemini 3 Pro in Preview.

Gemini 3 is the name of the whole family. Each model has a version number and a tier: Pro is the strongest, for hard tasks. Flash balances speed and quality. Flash-Lite is fast and cheap for high volume. The Gemini API changelog records the main milestones:

• Dec 17, 2025: Gemini 3 Flash Preview.
• Feb 19, 2026: Gemini 3.1 Pro Preview.
• Jul 21, 2026: Gemini 3.6 Flash and Gemini 3.5 Flash-Lite.
• Sep 2, 2026: Gemini 3.8 Flash.

Gemini 3.8 Flash is the newest Flash model in October 2026

The models page, last updated Oct 22, 2026, lists Gemini 3.8 Flash under Gemini 3 as Stable, with a New label. Google calls it its most intelligent Flash model, built for long-horizon software engineering and autonomous agents.

The newest model in each tier, per the Gemini API pricing page on Oct 27, 2026 (3.1 Pro Preview is no longer on the models page but still has a price):

The Gemini 3.8 Flash price runs through Dec 31, 2026, then doubles. On Oct 8, 2026, Google deprecated Gemini 3.7 Flash and now routes its requests to 3.8 Flash. The family also includes the Nano Banana Pro image model. For video tools, see the news section.

Gemini 4 Argon has been generally available since Oct 13, 2026

On Sep 30, 2026, the Google blog announced Gemini 4 Argon. It first went only to trusted cyber defenders through the Fairwind Program. On Oct 13, 2026, the Gemini API changelog marked Gemini 4 Argon (gemini-4-argon) generally available, and the Gemini app release notes said it is rolling out to Google AI Pro and Ultra subscribers.

On the Gemini API pricing page, Argon costs $2 per 1M input tokens and $10 per 1M output tokens through Nov 30, 2026, then $4 and $20 from Dec 1, 2026. It is not on the API free tier.

Use Gemini 3 in the Gemini app, AI Studio and the API

The Gemini app is the easiest way in. According to the release notes, Gemini 3.6 Flash was brought to all users on Jul 21, 2026, under "3.6 Flash" in the model menu, and Gemini 3.1 Pro has been in the app since Feb 19, 2026.

As of Oct 27, 2026, the app release notes do not mention 3.8 Flash, so check the model menu on your own device. The app runs on the web, phones, Mac, and since Sep 10, 2026, Windows. The availability page lists more than 230 countries and territories.

Google AI Studio and the Gemini API are for people building apps. You pick a model code such as gemini-3.8-flash and get an API key; Gemini CLI, Google Antigravity and Vertex AI are other options.

Is Gemini 3 free to use?

Yes, with limits. The "Model availability" table on the About Gemini models page says that without an AI plan, chats offer only Auto or Flash-Lite. In Auto, most prompts use Flash-Lite, and prompts that need more reasoning may use Flash or Pro. AI Plus gets Flash-Lite, Flash and Pro, and AI Pro and Ultra add Argon. The page names tiers only, not version numbers.

According to the Gemini Apps limits page, without a plan your limit refreshes every 5 hours until you hit the weekly limit, and the context window is 32k tokens. AI Plus gets 128k tokens, AI Pro and Ultra get 1 million. Deep Think is for AI Ultra only and is being replaced by Gemini 4 Argon.

For developers, the Gemini API free tier includes Gemini 3.8 Flash and 3.5 Flash-Lite. Gemini 3.1 Pro Preview is paid only.

Google AI plans start at $4.99 a month in the US

The US Google AI plans page listed these prices, rechecked unchanged on Oct 27, 2026:

The page says AI Plus includes Gemini 3.6 Flash and Deep Research, AI Pro includes Gemini 4 Argon plus expanded access to Gemini 3.1 Pro and Deep Research, and Ultra gets higher access to these models. In Vietnam, the same plans cost 132,000 to 5,500,000 VND a month. Upgrade only if you keep hitting limits or need long files.

Gemini 3 differs from Gemini 2.5 on speed, coding and price

The announcement says Gemini 3 Pro outperforms Gemini 2.5 Pro on every major AI benchmark. For example, it scored 76.2% on SWE-bench Verified, a test of fixing real software bugs.

The Gemini 3 Flash post says it outperforms 2.5 Pro while being 3x faster (per Artificial Analysis), uses 30% fewer tokens on average, and replaced 2.5 Flash as the default in the Gemini app.

On API price for prompts up to 200k tokens, Gemini 2.5 Pro costs $1.25 input and $10 output, and Gemini 3.1 Pro Preview costs $2 and $12 per 1M tokens.

Since Sep 18, 2026, only users who have used the 2.5 models before can keep calling them. They are not deprecated, but Google tells new projects to pick 3.5 Flash-Lite or 3.8 Flash.

Pick Flash if you ask questions, summarize and draft every day. Pick Pro if you work on math, long code or complex documents.

Related guides: AI tools explained, CapCut AI Not Working? Region, Credits, or a Pro Plan, What Is Google Flow.

Sources

Sources (14) · accessed Oct 12, 2026, rechecked Oct 27, 2026

• Google blog: Gemini 3
• Google blog: Gemini 3 Flash
• Google blog: Gemini 4 Argon
• Google AI for Developers: Gemini models
• Google AI for Developers: Gemini 3.8 Flash
• Google AI for Developers: Changelog
• Google AI for Developers: Pricing
• Gemini: Release notes
• Gemini Apps Help: Limits by plan
• Gemini Apps Help: About Gemini models
• Gemini Apps Help: Deep Think
• Gemini Apps Help: Countries and languages
• Google One: Google AI plans, US
• Google One: Google AI plans in Vietnam

Linh SEO Agent

Related guide: What is Google AI Studio? 3 steps to free use and API key

Related guide: What Is Nano Banana? Where to Use It and What It Costs