Faster and cheaper model takes center stage
Google has released Gemini 3 Flash, a fast and cost-efficient artificial intelligence model designed to compete directly with OpenAI’s latest offerings. Based on the Gemini 3 model unveiled last month, Gemini 3 Flash is now the default model in the Gemini app and in AI Mode within Google Search.
The launch comes just six months after Gemini 2.5 Flash and represents a major performance leap. Google says the new model significantly outperforms its predecessor and, in some benchmarks, rivals frontier models such as Gemini 3 Pro and GPT-5.2.
Benchmark results highlight performance gains
On Humanity’s Last Exam, a benchmark testing broad domain expertise, Gemini 3 Flash scored 33.7% without tool use. By comparison, Gemini 3 Pro reached 37.5%, GPT-5.2 scored 34.5%, and Gemini 2.5 Flash achieved just 11%.
The model also topped the MMMU-Pro multimodality and reasoning benchmark with an 81.2% score, outperforming all major competitors in that category.
Default rollout for consumers
Gemini 3 Flash is now live globally as the default model in the Gemini app, replacing Gemini 2.5 Flash. Users can still manually select Gemini 3 Pro for tasks such as advanced math and coding.
Google says the model excels at multimodal understanding. Users can upload images, videos, sketches, or audio and receive context-aware responses, such as sports tips from a short video, guesses about drawings, or analysis of audio recordings. The model also produces more visual answers, including tables and images.
Search, images and prototyping
Alongside Flash, Gemini 3 Pro is now available to all U.S. users in Search. Google is also expanding access to its Nano Banana Pro image model. Within the Gemini app, users can create basic app prototypes directly through prompts using the new model.
Enterprise and developer adoption
Google said companies including JetBrains, Figma, Cursor, Harvey, and Latitude are already using Gemini 3 Flash through Vertex AI and Gemini Enterprise. Developers can access the model in preview form via the API and through Antigravity, Google’s recently launched coding tool.
Gemini 3 Pro scores 78% on the SWE-bench verified coding benchmark, second only to GPT-5.2. Google positions Flash as a “workhorse” model suited for video analysis, data extraction, visual Q&A, and high-volume workflows.
Pricing and competitive context
Gemini 3 Flash is priced at $0.50 per million input tokens and $3.00 per million output tokens, slightly higher than Gemini 2.5 Flash. However, Google says the model is three times faster than Gemini 2.5 Pro and uses 30% fewer tokens for reasoning tasks, potentially lowering overall costs.
Google revealed it now processes more than one trillion tokens per day via its API, amid an intensifying AI rivalry with OpenAI. The release follows reports that OpenAI accelerated its own launches after seeing ChatGPT traffic dip as Google’s consumer share increased.