Gemini 3 Flash Now Becomes Google’s Default AI Model
• Gemini 3 Flash is now the default model in the Gemini app and AI mode for Google Search. • The model matches or beats frontier models on several benchmarks while running up to three times faster. • Available via Vertex AI, Gemini Enterprise, API preview and the Antigravity coding tool; adopted by JetBrains, Figma and others. • Pricing set at $0.50 per 1M input tokens and $3.00 per 1M output tokens, with token-efficiency gains for many tasks.
What’s changing
Google announced Gemini 3 Flash as a fast, lower-cost version of Gemini 3 and set it as the default model in the Gemini app worldwide. The company also flipped Gemini 3 Flash into AI mode for Search, replacing Gemini 2.5 Flash for most consumers.
Users can still pick Gemini 3 Pro from the model picker when they need stronger math or coding capabilities. Google also expanded access to Gemini 3 Pro in the U.S. and rolled out the Nano Banana Pro image model to more search users.
Performance and benchmarks
Google says Gemini 3 Flash outperforms the Gemini 2.5 Flash by a wide margin and reaches parity with top-tier models in many tasks. On Humanity’s Last Exam it scored 33.7% without tool use; Gemini 3 Pro scored 37.5% and GPT-5.2 scored 34.5% on the same test.
On multimodal reasoning (MMMU-Pro) Gemini 3 Flash led the field with an 81.2% score. The company also highlights that Gemini 3 Pro scores 78% on SWE-bench for coding tasks, second only to GPT-5.2 in Google’s cited comparisons.
Speed and efficiency
Google says Gemini 3 Flash runs about three times faster than its predecessor and uses roughly 30% fewer tokens on thinking tasks versus Gemini 2.5 Pro. That combination of speed and token efficiency targets high-volume, repeatable workflows.
Availability for developers and enterprises
Gemini 3 Flash is available through Vertex AI and Gemini Enterprise and is offered in a preview via Google’s API and the Antigravity coding tool. Google named early adopters including JetBrains, Figma, Cursor, Harvey and Latitude.
Google also reported processing more than 1 trillion tokens per day across its API since Gemini 3’s release, underscoring heavy production use.
Pricing and practical impact
Google priced Gemini 3 Flash at $0.50 per 1 million input tokens and $3.00 per 1 million output tokens. That compares with Gemini Flash 2.5’s $0.30/$2.50 rates but aims to offset higher per-token cost with fewer required tokens and faster throughput.
“We really position Flash as more of your workhorse model... Flash is just a much cheaper offering from an input and output price perspective,” said Tulsee Doshi, senior director & head of Product for Gemini Models, in a briefing with TechCrunch.
Why it matters
Making Gemini 3 Flash the default shifts Google’s consumer footprint toward a model optimized for speed and multimodal tasks like video analysis, visual Q&A and app prototyping inside the Gemini app. For enterprises and developers, the model targets bulk tasks and fast, repeatable pipelines.
The release also intensifies competition with OpenAI’s GPT-5.2, continuing a rapid cycle of model releases and benchmark-driven comparisons across the industry.
