As of Nov 2026, Google's Gemini API models page leads with Gemini 4 Argon, its flagship reasoning model with a 1M token context window, followed by the Gemini 3.x Flash and Flash-Lite models, voice and speech models, the Nano Banana image models and Gemini Omni Flash for video. Google's own advice for new projects is Gemini 3.8 Flash or Gemini 3.5 Flash-Lite, and the Stable and Preview labels tell you which models are safe to build on.
One thing first: this is the API view, meaning the model names developers put into code. The Gemini app may present its choices differently, so treat this page as the map you'd hand to whoever is building for you.
The Gemini line-up on one page
Here's how the models page groups things as of Nov 2026, with the endpoint string a developer would use and the label Google gives each one.
| Job | Model | Endpoint | Label |
|---|---|---|---|
| Hardest reasoning, very long documents | Gemini 4 Argon | gemini-4-argon | New, Stable |
| Everyday work, agents, coding | Gemini 3.8 Flash | gemini-3.8-flash | New, Stable |
| Previous-generation Flash | Gemini 3.6 Flash | gemini-3.6-flash | Stable |
| High volume at low cost | Gemini 3.5 Flash-Lite | gemini-3.5-flash-lite | Stable |
| Low cost | Gemini 3.1 Flash-Lite | gemini-3.1-flash-lite | Stable |
| Live voice agents | Gemini 3.8 Live | gemini-3.8-live | New, Stable |
| Voice with deeper reasoning | Gemini 3.8 Live Extended Thinking | gemini-3.8-live-extended-thinking | New, Stable |
| Text to speech | Gemini 3.8 Flash TTS | gemini-3.8-flash-tts | New, Stable |
| Transcription | Gemini 3.5 Transcribe | gemini-3.5-transcribe | New, Stable |
| Image generation and editing | Nano Banana 2.1 | gemini-nano-banana-2.1 | New, Stable |
| Studio-quality images | Nano Banana Pro | gemini-3-pro-image | Stable |
| Video | Gemini Omni Flash | gemini-omni-1.1-flash | Stable |
| Live translation | Gemini 3.5 Live Translate | gemini-3.5-live-translate-preview | New, Preview |
Gemini 4 Argon, the flagship
Google describes Gemini 4 Argon as its flagship reasoning model with a unified 1M context window and long decodes up to 1M tokens. In plain words, it can take in a very large amount of material and write a very long answer back. It's the only model in the Gemini 4 group so far, and it carries both the New and Stable labels. I've written up how to use Gemini 4 Argon in practice if you want to see it at work.
Flash and Flash-Lite, the everyday workhorses
Gemini 3.8 Flash is described as Google's most intelligent Flash model, built for long-horizon software engineering, autonomous agents and complex enterprise workflows. Gemini 3.6 Flash is now the previous-generation Flash model, still Stable, balancing speed and multimodal skills across everyday tasks.
Below that sit the budget models. Gemini 3.5 Flash-Lite is the fastest, most cost-effective 3.5 model for high-throughput work, and Gemini 3.1 Flash-Lite is the other low-cost option. Two older names still show in Preview: Gemini 3 Flash, and Gemini 3.1 Pro, which appears only as a preview endpoint. If you're catching up from the last generation, my Gemini 3 explainer covers where these names came from.
Voice, speech and transcription
This is the busiest corner of the page. Gemini 3.8 Live is the default Live API model for most low-latency voice agents, without reasoning delays. Gemini 3.8 Live Extended Thinking is the one to use when a voice conversation needs more background reasoning.
For reading text aloud, Gemini 3.8 Flash TTS is the flagship, with studio-grade voice fidelity, expressive acting, Voice design and Voice replication across 130 languages. Gemini 3.8 Flash-Lite TTS is the fast, cost-efficient one for high-volume production, across 101 languages. Gemini 3.5 Transcribe turns speech into text with speaker diarization, word-level timestamps and custom vocabulary biasing, which is handy for meeting notes or podcast transcripts.
Watch the labels here. Gemini 3.5 Live Translate, which handles real-time speech to speech translation in 70+ languages, is still Preview. And Gemini 3.1 Flash Live and Gemini 3.1 Flash TTS are marked as legacy preview models, with Google recommending a move to the 3.8 versions.
Images and video: Nano Banana and Omni
Google's image models all carry the Nano Banana name. Nano Banana 2.1 is the newest, with improved visual quality, typography and multi-turn consistency. Nano Banana 2 is built for speed and high-volume use, Nano Banana 2 Lite is the ultra-low latency, cost-effective option, and Nano Banana Pro is described as a professional design engine with a reasoning core for studio-quality 4K visuals, complex layouts and precise text rendering. The older Imagen 4 model is now listed as shut down.
For video, Gemini Omni Flash handles video generation, editing, keyframe interpolation and extension with native audio, under the endpoint gemini-omni-1.1-flash. My walkthrough on using Gemini Omni shows what that looks like in practice.
The page also lists models most business owners won't touch directly: Lyria 3.5 for full-length music, Gemini Deep Research and the Antigravity Agent as preview agent models, and Gemini Embedding 2 for search systems.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Stable, Preview, legacy and shut down: reading the labels
These labels are the most useful thing on the page, because they tell you how much you can rely on a model.
- Stable points to a specific model that usually doesn't change. Google says most production apps should use one.
- Preview models may be used for production, but typically have billing enabled, may have tighter rate limits, and will be deprecated with at least 2 weeks' notice.
- Latest is an alias that gets hot-swapped with every new release, with a 2-week email notice before breaking changes.
- Experimental models typically aren't suitable for production, and their endpoints can change.
There's also a quiet catch on the Gemini 2.5 models. Google is limiting access to them to users who actively used them in the past. They're not deprecated and are still served, but for any new project Google says to use 3.5 Flash-Lite or 3.8 Flash. Further down, Gemini 2.0 Flash, Gemini 2.0 Flash-Lite and Gemini 3 Pro Preview are all marked as shut down.
What I'd pick for a small business
I'd start on Gemini 3.8 Flash for most jobs, or 3.5 Flash-Lite for big repetitive batches, and only move up to Argon if your own test on real work falls short. Skip Preview models for anything customers see: two weeks' notice isn't much when a chatbot or a booking flow depends on it. Prices aren't on the models page, so check Google's pricing page before you choose, and if you're testing on a budget, here's what the free Gemini API tier gives you. Names change quickly, so treat this as the picture as of Nov 2026.
If you want a plain-English view of which AI model fits which job in your business, that's the kind of AI enablement I help with. I run AI training and workshops for teams, and help businesses with AI enablement and workflow automation, so the busywork runs itself and people get time back. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.
Frequently asked questions
What is the newest Gemini model?
As of Nov 2026, the Gemini API models page leads with Gemini 4 Argon, Google's flagship reasoning model with a 1M token context window, marked New and Stable.
Which Gemini model should I use for everyday tasks?
Google's models page recommends Gemini 3.8 Flash or Gemini 3.5 Flash-Lite for new projects. Flash-Lite is the fastest, most cost-effective 3.5 model.
What's the difference between Stable and Preview Gemini models?
Stable models usually don't change and suit production apps. Preview models may have tighter rate limits and can be deprecated with at least 2 weeks' notice.
Can I still use Gemini 2.5 models?
Only if you've used them before. Google is limiting access to existing users, though the models aren't deprecated and are still served.
What is Nano Banana?
It's the name of Google's image generation and editing models in the Gemini API, from Nano Banana 2 Lite up to Nano Banana Pro. Nano Banana 2.1 is the newest.
