Google has unveiled Gemini 3 Flash, a faster and more efficient AI model that brings Gemini 3-level reasoning to developers, enterprises and everyday users at a lower cost.
Google has expanded its Gemini 3 model line-up with Gemini 3 Flash, a new version designed to deliver “frontier” reasoning while keeping latency and costs low.
In a blog post, Google said Gemini 3 Flash combines the “Pro-grade” reasoning it introduced with Gemini 3 Pro, but tuned for faster responses and higher efficiency – a pairing the company is pitching for both everyday use and more advanced “agentic” workflows that require rapid iteration.
Google is also leaning heavily on benchmark results to frame the upgrade.
The company said Gemini 3 Flash scores 90.4 per cent on GPQA Diamond, 33.7 per cent on Humanity’s Last Exam (without tools), and 81.2 per cent on MMMU Pro, while using fewer tokens on average than Gemini 2.5 Pro under typical traffic.
Google has claimed this efficiency allows the model to scale without sacrificing reasoning quality.
For developers, Gemini 3 Flash is arriving in preview through the Gemini API in Google AI Studio, Gemini CLI, and Google’s agent-focused tooling, alongside availability for enterprises via Vertex AI and Gemini Enterprise.
Google has said the model is particularly suited to coding agents, multimodal analysis and real-time applications that demand both speed and accuracy.
Pricing is another key part of the pitch, as Google lists Gemini 3 Flash at $0.50 per million input tokens and $3 per million output tokens, positioning it as a cheaper on-ramp to Gemini 3 performance for high-volume applications.
For consumers, Google has said Gemini 3 Flash is becoming the default model in the Gemini app, replacing 2.5 Flash, and is also rolling into AI Mode in Search – a sign that Gemini 3 is moving from “latest model” to baseline infrastructure across Google’s products.
Google unveils new Gemini 3 model
