Google launches Gemini 3 Flash, makes it the default model in the Gemini app - TechCrunch
Google has launched Gemini 3 Flash—a faster, lower-cost multimodal model—and made it the default in the Gemini app and AI mode in Search. Positioned as a high-speed "workhorse," Gemini 3 Flash aims to undercut rivals on cost while matching or beating them on key benchmarks, and it replaces Gemini 2.5 Flash as the default. Users can still switch to the Pro model for math and coding tasks.
Why it matters
- Default everywhere: Gemini 3 Flash is now the default model in the Gemini app globally and in AI mode in Search.
- Speed and value: It’s designed for fast, repeatable workflows and bulk tasks while maintaining strong reasoning and multimodal performance.
Performance and benchmarks
- Humanity’s Last Exam (no tool use): Gemini 3 Flash scores 33.7%, vs. Gemini 3 Pro at 37.5%, Gemini 2.5 Flash at 11%, and GPT-5.2 at 34.5%.
- MMMU-Pro (multimodality and reasoning): Gemini 3 Flash leads with 81.2%, outscoring all competitors.
- Coding (SWE-bench verified): Gemini 3 Pro hits 78%, second only to GPT-5.2.
What you can do with Gemini 3 Flash
- Multimodal understanding: Upload short videos for coaching tips (e.g., pickleball), submit sketches for recognition, or analyze audio to generate summaries and quizzes.
- More visual answers: Expect richer responses that can include images and tables.
- Rapid prototyping: Build app prototypes directly in the Gemini app using prompt-based workflows.
Consumer availability
- Gemini 3 Flash is the default model in the Gemini app worldwide.
- Users can switch to Gemini 3 Pro via the model picker for math/coding use cases.
- Gemini 3 Pro is rolling out to everyone in the U.S. for Search.
- More U.S. users can now access the Nano Banana Pro image model in Search.
Enterprise and developer access
- Early adopters include JetBrains, Figma, Cursor, Harvey, and Latitude.
- Available through Vertex AI and Gemini Enterprise.
- Developer preview via API and in Antigravity, Google’s new coding tool.
Pricing and efficiency
- Gemini 3 Flash: $0.50 per 1M input tokens, $3.00 per 1M output tokens.
- Previous 2.5 Flash: $0.30 per 1M input; $2.50 per 1M output.
- Despite a slight price bump, Google says Gemini 3 Flash outperforms Gemini 2.5 Pro while running 3x faster—and uses ~30% fewer tokens on average for “thinking” tasks, which can lower overall costs.
Industry context
- Google reports processing over 1 trillion tokens per day on its API following the Gemini 3 launch.
- Amid rising competition, OpenAI released GPT-5.2 and a new image generator, citing 8x ChatGPT message growth since November 2024.
- Google frames the race as a positive driver of innovation and better benchmarks across the industry.
Bottom line
Gemini 3 Flash becomes Google’s default consumer model with standout multimodal scores, faster performance, and lower total cost for many workflows—aimed squarely at bulk, repeatable tasks while preserving strong reasoning chops.