Gemini Blog - One API Access 500+ AI Models

Google unveils Gemini 3.1 Flash-Lite — a fast, low-cost LLM

Google introduced Gemini 3.1 Flash-Lite, the newest member of the Gemini 3 family designed specifically as a high-throughput, low-latency, cost-efficient engine for developer and enterprise workloads. Google positions Flash-Lite as the “fastest and most cost-efficient” model in the Gemini 3 line: a lightweight variant that aims to deliver streaming interactions, large-scale background processing, and high-frequency production tasks (for example, translation, extraction, UI generation, and large-volume classification) at a much lower price point than its Pro counterparts

Gemini 3.1 Flash is Coming Soon: What it is

Gemini 3.1 Flash—the ultra-low-latency, image-capable member of the Gemini 3.1 family—is being rolled out across Google’s consumer and developer surfaces. Gemini 3.1 Flash builds close the gap between reasoning quality and responsiveness. For image tasks, the Flash Image variant improves on text rendering within images and maintains coherent identities for multiple characters and objects across a workflow — a common pain point for earlier image models.

Gemini 3.1 Pro is Now Live on CometAPI: What it is and how to access

The Gemini 3.1 Pro is now available on CometAPI, and you can start using it through CometAPI's services—at a more affordable launch price than the official price. CometAPI already exposes the Gemini 3 family and provides an OpenAI-compatible path to call those models from a single unified gateway; that makes it quick to experiment with Gemini models using existing OpenAI SDKs

Google Gemini 3.5(Snow Bunny) Leaked: All you need to know

Google is quietly testing a new internal iteration of its Gemini family — reported variously as “Gemini 3.5” and by the intriguing internal codename “Snow Bunny.” Codenamed "Snow Bunny," this internal checkpoint has reportedly shattered existing benchmarks, demonstrating an unprecedented ability to generate entire software applications—up to 3,000 lines of functional code—in a single prompt.

Shopping on Google: How do you use Google’s AI shopping as Merchants

Google has reworked its shopping experience around generative AI and the Gemini family of models. For consumers, the shift promises conversational product discovery, AI-generated comparison briefs, and — where available — automated “agentic” checkout that can buy on your behalf when preconditions are met. For merchants and developers, the new surface combines two sets of APIs (shopping / merchant APIs and Google’s GenAI / Gemini APIs) and requires updated feed practices, privacy controls, and technical integration.

Gemini API: what file sizes & input methods are supported now?

On January 12, 2026 Google published a developer update to the Gemini API that changes how you get files into the model and how large those files can be. In short: Gemini now fetches files directly from external links and cloud storage (so you don't always have to upload them), and the inline file size limit has been raised substantially. These updates remove friction for real-world apps that already store media or documents in cloud buckets, and make short prototyping and production workflows faster and cheaper.

Mar 30, 2026

Gemini 3 Flash vs Gemini 3 Pro: Price, Speed & Reasoning

Gemini 3 Flash — optimized for raw throughput, low latency, and cost efficiency — and Gemini 3 Pro — optimized for the deepest multimodal reasoning, largest context windows and highest benchmark ceilings. In practical terms, Flash is designed to shift the “productive-flow” frontier for high-frequency developer and interactive applications; Pro is designed to maximize single-query intelligence and handle very large or complex multimodal inputs. The tradeoffs are straightforward and measurable: Flash delivers substantially lower latency and materially lower per-token costs while keeping much of Gemini 3’s reasoning ability; Pro delivers the highest benchmark scores, the most advanced modes (e.g., Deep Think), and larger safety-guarded capabilities at higher cost and latency.

Is Gemini 3 Pro Good for Coding? A 2026 Reality-Check and Practical Guide

Google’s Gemini 3 Pro arrived as a headline-grabbing multimodal model that Google positions as a major step forward in reasoning, agentic workflows, and coding assistance. In this long-form piece I note to answer one clear question: Is Gemini 3 Pro good for coding? Short answer: Yes — with important caveats.