Gemini 2.5 Pro is Google's most advanced model in the 2.5 family, built for complex tasks with deep reasoning and coding, and it's still running. As of Nov 2026, Google's Gemini API models page says the 2.5 models aren't deprecated, but access is limited to people who've actively used them before, and new projects should use the latest models: 3.5 Flash-Lite or 3.8 Flash.
That's the short version. The rest of this page explains what Google's note means for a business that relies on an AI tool built on Gemini, how to find out whether yours is one of them, and what I'd do about it.
What Google says Gemini 2.5 Pro is for
Google's description on its models page is one line: "Our most advanced model for complex tasks, featuring deep reasoning and coding capabilities in the 2.5 family." Its API code, the string a developer types to call it, is gemini-2.5-pro.
One thing to know before you read further. That page is developer documentation for the Gemini API, part of Google AI for Developers. It lists the model codes that apps and tools call behind the scenes. It isn't a guide to the Gemini app on your phone, and it doesn't say which models the app shows you. So when I talk about access below, I mean access through the API, which is how most third-party tools, plugins and in-house automations reach Gemini.
Can you still use it?
Yes, if you were already using it. Google has put the same note above Gemini 2.5 Pro, 2.5 Flash and 2.5 Flash-Lite, and it says three things:
- Access is now limited to people who've actively used the 2.5 models in the past, to keep performance reliable for everyone.
- The models aren't deprecated and will keep being served through the API until further notice.
- For any new project, Google tells you to use its latest models, 3.5 Flash-Lite or 3.8 Flash.
Google gives a reason too: it wants enough capacity for both ongoing legacy workflows and new applications. In plain words, the 2.5 models are being kept alive for the people who already built on them, not offered to newcomers.
That's the part most people miss when they search for this model. It hasn't been switched off, so a tool that already runs on it should keep working. But a brand new account, or a developer starting a fresh project, may not be able to pick it at all. If someone tells you 2.5 Pro has been retired, that's not what Google's page says. If someone tells you to build your next project on it, that's not what the page says either.
The 2.5 family in one table
Here's the 2.5 line-up as Google lists it, with one newer model at the bottom for comparison. The descriptions are Google's, shortened.
| Model | What Google says it's for | API code |
|---|---|---|
| Gemini 2.5 Pro | Most advanced 2.5 model, for complex tasks with deep reasoning and coding | gemini-2.5-pro |
| Gemini 2.5 Pro TTS | High-fidelity speech for structured work like podcasts and audiobooks | gemini-2.5-pro-preview-tts |
| Gemini 2.5 Flash | Best price-performance for low-latency, high-volume tasks that need reasoning | gemini-2.5-flash |
| Gemini 2.5 Flash-Lite | Fastest and most budget-friendly multimodal model in the 2.5 family | gemini-2.5-flash-lite |
| Nano Banana | Native image generation and editing for fast, creative workflows | gemini-2.5-flash-image |
| Gemini 3.1 Pro | Listed in the Gemini 3 family, still marked preview | gemini-3.1-pro-preview |
Notice that Gemini 2.5 Pro TTS is a separate model. It turns text into speech; it isn't a voice setting on 2.5 Pro. If a tool that reads scripts aloud mentions "2.5 Pro", ask which of the two it means. Nano Banana is the same story: its code starts with gemini-2.5, but it's an image model. The family also includes 2.5 Flash Live, for real-time voice agents, and 2.5 Flash TTS.
What sits above it now
The 2.5 models are no longer at the top of Google's page. As of Nov 2026 it leads with Gemini 4 Argon, which Google calls its flagship reasoning model, then the Gemini 3 family, including 3.8 Flash, described as Google's most intelligent Flash model, plus 3.6 Flash and 3.5 Flash-Lite. If you want the newer generation explained, I've written about what Gemini 3 is and what changed, and my guide to the full Gemini model line-up covers the rest, so I won't re-list them here.
What I won't do is tell you 2.5 Pro is better or worse than any of these. Google's page doesn't rank them against each other, and the answer depends on your task. The models page doesn't cover cost either; for that, see the Gemini API pricing guide on this site and check Google's own pricing page for the current figures.
Stable, preview and latest: the labels that decide your warning time
The section of Google's page that matters most to a business owner is near the bottom, under model version name patterns. It explains what each kind of model code means:
- Stable: points to a specific model that usually doesn't change. Google says most production apps should use one. gemini-2.5-pro is a stable code.
- Preview: can be used in production, but may have tighter rate limits and will be deprecated with at least 2 weeks' notice.
- Latest: an alias such as gemini-flash-latest that gets hot-swapped with every new release. For breaking changes, Google sends a 2-week notice by email before the model behind it changes.
- Experimental: not usually suitable for production, with more restrictive rate limits and endpoints that can change.
Why should you care? Because a tool on a latest alias can change its behaviour with two weeks' warning, and that email goes to whoever owns the API account, which may be your developer, not you. A preview model can go with the same short notice. The page's own previous models section shows how that ends: Gemini 2.0 Flash, Gemini 2.0 Flash-Lite and Gemini 3 Pro Preview are all marked shut down, and so is a computer use preview whose code starts with gemini-2.5. So a 2.5 in the name doesn't mean a model is covered by Google's note. Only the models listed under that note are.
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.
How to check whether your tools rely on it
Most business owners never touch the Gemini API directly. You meet it through a writing assistant, a chatbot on your site, a reporting tool or an automation someone built for you. Here's how I'd find out where you stand:
- List the tools that use Gemini. Check settings pages, help docs and invoices for the words Gemini or Google AI.
- Ask for the exact model code. "Gemini" isn't enough. You want the string, such as gemini-2.5-pro or gemini-3.8-flash.
- Check that code against Google's models page. See which section it sits in, whether it's stable, preview or latest, and whether it's under the 2.5 note or marked shut down.
- Ask what happens if access ends. Who gets Google's emails, how much notice your vendor will pass on, and whether there's a fallback model.
- Plan the move for anything new. If a tool or workflow is still being built, ask for it to be built on one of the latest models Google names instead.
If you don't know who to ask, start with whoever set the tool up. A good developer or vendor will answer all five in one email, and if they can't tell you which model code they use, that tells you something too.
My verdict on Gemini 2.5 Pro
I'd leave any workflow that already runs well on 2.5 Pro alone until Google or your vendor sends a notice that says otherwise. Google's page says it's not deprecated, and rebuilding something that works just because a newer model exists is how teams lose a fortnight.
I'd skip it for anything new. Google tells new projects to use its latest models, access is limited to past users anyway, and you don't want to start a build on a model that's being kept going for legacy work. Before you switch an existing workflow, test the newer model on one task you do every week, compare the output side by side, and only then move the rest.
Google updates this page often (it was last updated on 28 October 2026 when I checked), so treat everything here as as of Nov 2026 and check Google's own models page and its Gemini deprecations page for the latest.
If your team's AI tools sit on older Gemini models and you'd like a calm plan for moving them, 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
Is Gemini 2.5 Pro deprecated?
No. As of Nov 2026 Google's Gemini API models page says the 2.5 models aren't deprecated and will keep being served through the API until further notice. Access is limited to people who've actively used them before.
Can I use Gemini 2.5 Pro for a new project?
Google says not to. Its note tells new projects to use its latest models, 3.5 Flash-Lite or 3.8 Flash, and access to the 2.5 models is limited to past users.
What is the API code for Gemini 2.5 Pro?
The endpoint is gemini-2.5-pro. The speech model, Gemini 2.5 Pro TTS, is separate and uses gemini-2.5-pro-preview-tts.
Does Google's models page cover the Gemini app?
No. It's documentation for the Gemini API, the route developers and third-party tools use. It doesn't say which models the Gemini app shows you.
How much notice will I get if a model changes?
Google says preview models are deprecated with at least 2 weeks' notice, and a latest alias gets a 2-week email notice before a breaking change. That email goes to whoever owns the API account.
