Gemini 3 Flash is rolling out globally in Google Search

1. Integration into Google Search

• AI Mode: Gemini 3 Flash is becoming the default model for AI Mode in Google Search (often appearing as AI Overviews).
• Impact: This upgrade is designed to significantly reduce latency. It allows Search to handle more complex, multi-layered queries (like planning trips or comparing products with specific constraints) much faster than previous models, without sacrificing the quality of the reasoning.

2. What is Gemini 3 Flash?

Google positions this model as “frontier intelligence built for speed.” It is designed to offer a specific balance:

• Speed: It matches the low latency and cost-efficiency typical of the “Flash” series (optimized for high-frequency tasks).
• Reasoning: Unlike previous lightweight models, it claims to offer reasoning capabilities comparable to the heavier “Pro” models (specifically Gemini 3 Pro).
• Multimodal: It retains strong capabilities in processing text, images, and video simultaneously.

3. Broader Availability

In addition to Search, this rollout covers other major platforms:

• The Gemini App: It has replaced previous versions as the default model for free users in the Gemini app globally.
• For Developers: It is available via the Gemini API, Google AI Studio, and Vertex AI for enterprise use.

Here is the pricing comparison between the newly released Gemini 3 Flash and the older Gemini 1.5 Flash.

Pricing Comparison (Per 1 Million Tokens)

Key Differences

• Price Increase: Gemini 3 Flash is significantly more expensive (roughly 6x higher for input and 10x higher for output) than the original 1.5 Flash.
• Performance Shift: The price difference reflects a major shift in capability. While 1.5 Flash was a “lightweight” model designed for simple tasks, Gemini 3 Flash is a “frontier” model. It effectively delivers the high-level reasoning of a “Pro” model (outperforming even the recent Gemini 2.5 Pro) but at the speed of a Flash model.
• Pricing Structure: Gemini 3 Flash uses a flat pricing model, meaning you pay the same rate regardless of how long your prompt is. In contrast, Gemini 1.5 Flash used tiered pricing where costs doubled if your prompt exceeded 128k tokens.