Direct answer
“Gemini 3 Pro” is the product name; gemini-3-pro-preview was the original Gemini API model ID. That model was shut down on March 9, 2026, so don’t use it for a new production integration. Google’s current documentation lists Gemini 3.1 Pro Preview with the model code gemini-3.1-pro-preview. It is a preview model, not the same model as the retired Gemini 3 Pro Preview. (ai.google.dev)
| What you mean | Name or identifier | What to know |
|---|---|---|
| The original model’s display name | Gemini 3 Pro | A readable product name, not by itself an API ID. |
| The original Gemini API model | gemini-3-pro-preview |
Google lists it as shut down on March 9, 2026. Don’t use it for new requests. (ai.google.dev) |
| The documented Gemini 3.1 Pro Preview model | gemini-3.1-pro-preview |
Google lists this as the model code. It remains a preview model; confirm it is available for your project and channel before deployment. (ai.google.dev) |
Which identifier should you use?
If you are starting a new Google API integration and want the Pro model in the Gemini 3 series, use gemini-3.1-pro-preview only if preview access and terms fit your use case. Google describes it as designed for software engineering and agentic workflows involving precise tool use and multi-step execution. That is a documented positioning, not a guarantee of results for your application—test it on your own tasks. (ai.google.dev)
For Vertex AI, Google’s getting-started guide also specifies gemini-3.1-pro-preview and says it is available only on the global endpoint. The guide shows GOOGLE_CLOUD_LOCATION=global; a regional endpoint may not work for this model. (docs.cloud.google.com)
For a production workload that cannot depend on a preview model, don’t assume the word “Pro” means a stable production release. Check the model’s current lifecycle and terms in the exact Google API channel you plan to use, then evaluate an available generally available model if preview status is unsuitable. Google’s Gemini API models.list endpoint can return the models available through that API; use it to confirm availability for your API key rather than copying an old identifier from an article. (ai.google.dev)
Don’t confuse these names with API IDs
A third-party article published November 6, 2025 claimed the names gemini-3-pro-preview-11-2025 and gemini-3-pro-preview-11-2025-thinking. The official Google documentation captured here instead gives gemini-3-pro-preview for the original model and marks it shut down. The article’s longer strings therefore should not be treated as confirmed current Google model IDs. (ai.google.dev)
Likewise, Gemini 3 Pro Image is a separate image-generation product. A third-party platform calls it “Nano Banana Pro” and uses its own identifiers, including gemini-3-pro-image. Those platform-specific image identifiers are not the general-purpose Gemini Pro model ID. (aistudio.google.com)
In short: the original Gemini 3 Pro API ID was gemini-3-pro-preview, now shut down. The documented Pro model in the Gemini 3.1 series is gemini-3.1-pro-preview; for Vertex AI, use the global endpoint, and account for its preview status.
Sources
Gemini 3 Pro preview — Gemini API
Gemini 3.1 Pro preview — Gemini API