Key Points
- Google unveils Gemini 3 Flash, now the default model in the Gemini app and AI search.
- The model is designed to be faster and more affordable, competing head‑to‑head with OpenAI.
- Benchmarks show major performance gains, including an 81.2% score on MMMU‑Pro, leading all rivals.
- Gemini 3 Flash is rolling out globally for consumers, enterprises, and developers.
- Pricing starts at $0.50 per million input tokens, with Google claiming it is three times faster than the previous Pro model.
Google Introduces Its New High‑Speed, Low‑Cost AI Model
Google has officially launched Gemini 3 Flash, a streamlined AI model built for speed, affordability, and broad accessibility. Announced today, the model immediately becomes the default engine powering the Gemini app and Google’s AI search features.
Built on the architecture of last month’s Gemini 3 Pro, the new Flash model is positioned as Google’s answer to OpenAI’s latest offerings. According to TechCrunch, the rollout is global and marks a significant escalation in the ongoing AI competition.
Benchmark Results Reveal Major Performance Gains
Early benchmarks show Gemini 3 Flash delivering substantial improvements over its predecessor, Gemini 2.5 Flash, and performing competitively against top-tier models.
Key Benchmark Scores
- Humanity’s Last Exam: 33.7%
- Gemini 3 Pro: 37.5%
- OpenAI GPT‑5.2: 34.5%
- MMMU‑Pro (Multimodal Reasoning): 81.2%
- Highest score among all tested models
The strong multimodal performance highlights the model’s ability to interpret and reason across images, text, audio, and video—an increasingly critical capability in modern AI systems.
Consumer and Enterprise Rollout Begins
Gemini 3 Flash is now the default model for everyday users of the Gemini app, replacing the older Flash version. Users still have the option to switch to the Pro model for more complex tasks such as coding or advanced analysis.
The model supports a wide range of multimodal inputs:
- Upload a short video for sports technique feedback
- Draw a sketch for identification
- Analyze audio recordings
- Process mixed media queries
Enterprise Adoption Already Underway
Companies including JetBrains, Figma, and Cursor have already integrated Gemini 3 Flash into their workflows. Businesses can access the model through Vertex AI and Gemini Enterprise, while developers can experiment with it via API preview and Google’s coding tool, Antigravity.
Google reports that its APIs now process over one trillion tokens per day, underscoring the scale of demand for its AI infrastructure.
Pricing and Performance Position Gemini 3 Flash as a “Workhorse” Model
Google has priced the model aggressively:
- $0.50 per million input tokens
- $3.00 per million output tokens
The company claims Gemini 3 Flash not only outperforms the older Pro model in several areas but also operates three times faster, making it an attractive option for high‑volume applications.
What Happens Next?
With Gemini 3 Flash entering the market, Google is signaling a clear intent to challenge OpenAI’s dominance by offering a faster, cheaper, and highly capable alternative. As enterprises and developers begin adopting the model at scale, its impact on the AI landscape—and on competitive pricing—will become more apparent.
The AI arms race is accelerating, and Gemini 3 Flash positions Google as a formidable contender in the next phase of the industry’s evolution.
