Google prepares Gemini 3 AI to take on OpenAI's GPT-5
What's the story
Google's next big artificial intelligence (AI) model, Gemini 3, has been spotted in early tests within the company's AI Studio. The discovery comes as developers have reported sightings of the new model's references in A/B tests across the platform. Screenshots and code snippets shared on social media show "flash-preview" builds being tested with significant improvements in coding and image generation tasks.
Performance boost
Gemini 3 outpacing Claude 4.5 in early tests
The new model, Gemini 3, is already outpacing its competitor Claude 4.5 in handling vector graphics and structured outputs. This was reported by TestingCatalog and several developers on X. The news comes as Google Cloud's official account teased an October 9 livestream under the hashtag #GeminiAtWork, further fueling speculation that this could be the event where Gemini 3 is officially unveiled to the public.
AI showdown
Gemini 3 could be Google's response to GPT-5
Gemini 3 is expected to be Google's biggest AI update of 2025, with improvements in reasoning, coding accuracy, and creative generation. Analysts believe it could be the company's strongest response yet to OpenAI's GPT-5 and Anthropic's Claude 4.5. Both models have set high standards for AI performance. A more capable Gemini could help Google regain its position in generative AI, especially if integrated into Workspace, Chrome, and Android.
Developer advantage
What to expect from Gemini AI model
Gemini 3 is expected to offer faster, context-aware code generation for developers. It may also introduce an "Agent Mode" for browser-level automation. This could be a game-changer in the tech world, especially if Google confirms and launches the new model during Wednesday's #GeminiAtWork livestream. If so, we could see a gradual rollout through AI Studio and Workspace tools in the coming weeks.