Gemini 3 has been released and is progressively being made available to Gemini Code Assist users in VS Code and IntelliJ. Initially, Gemini 3 is only being made available in agent mode.
Once enabled, Gemini Code Assist tries to use Gemini 3 unless otherwise configured using the model selector (VS Code only). When Gemini Code Assist tries to use Gemini 3, it falls back to other models, such as Gemini 2.5 Pro, if capacity is unavailable or quotas are exceeded. Responses include a label when they are generated using Gemini 3.
Get Access to Gemini 3
| License / Subscription | Gemini 3 Availability |
|---|---|
| Google AI Ultra | Available now to all users in IntelliJ; Available soon to all users in VS Code |
| Google AI Pro | Join the waitlist |
| Gemini Code Assist Enterprise | Available soon to all users in the Preview release channel |
| Gemini Code Assist Standard | Join the waitlist |
| Gemini Code Assist for individuals | Join the waitlist |
| Google Developer Program | Join the waitlist |
Use Gemini 3 in Agent Mode
VS Code
Gemini 3 must be enabled in the Gemini CLI to be used in Agent Mode in VS Code.
Close VS Code.
Install Gemini CLI version 0.16 or later if you haven't already done so.
Run Gemini CLI after installation using the following command:
gemini
In the Gemini CLI, open settings using the following command:
/settings
Navigate to the Preview Features setting and enable it.
Reopen VS Code.
Once Gemini 3 is enabled for your license or subscription, and Preview Features are enabled in Gemini CLI, Gemini 3 works for Agent Mode in VS Code and falls back to other models, such as Gemini 2.5 Pro, if capacity is unavailable or quotas are exceeded.
IntelliJ
Once Gemini 3 is enabled for your license or subscription, Gemini 3 automatically works for Agent Mode in IntelliJ and falls back to other models, such as Gemini 2.5 Pro, if capacity is unavailable or quotas are exceeded.