DeeperML

Google AI has released Gemma 3, a new family of open-source AI models designed for efficient and on-device AI applications. Gemma 3 models are built with technology similar to Gemini 2.0, intended to run efficiently on a single GPU or TPU. The models are available in various sizes: 1B, 4B, 12B, and 27B parameters, with options for both pre-trained and instruction-tuned variants, allowing users to select the model that best fits their hardware and specific application needs.

Gemma 3 offers practical advantages including efficiency and portability. For example, the 27B version has demonstrated robust performance in evaluations while still being capable of running on a single GPU. The 4B, 12B, and 27B models are capable of processing both text and images, and supports more than 140 languages. The models have a context window of 128,000 tokens, making them well suited for tasks that require processing large amounts of information. Google has built safety protocols into Gemma 3, including a safety checker for images called ShieldGemma 2.

ImgSrc: storage.googlea

References :

• MarkTechPost: Google AI Releases Gemma 3: Lightweight Multimodal Open Models for Efficient and On‑Device AI
• The Official Google Blog: Introducing Gemma 3: The most capable model you can run on a single GPU or TPU
• AI News | VentureBeat: Google unveils open source Gemma 3 model with 128k context window
• AI News: Details on the launch of Gemma 3 open AI models by Google.
• The Verge: Google calls Gemma 3 the most powerful AI model you can run on one GPU
• Maginative: Google DeepMind’s Gemma 3 Brings Multimodal AI, 128K Context Window, and More
• TestingCatalog: Gemma 3 sets new benchmarks for open compact models with top score on LMarena
• AI & Machine Learning: Announcing Gemma 3 on Vertex AI
• Analytics Vidhya: Gemma 3 vs DeepSeek-R1: Is Google’s New 27B Model a Tough Competition to the 671B Giant?
• AI & Machine Learning: How to deploy serverless AI with Gemma 3 on Cloud Run
• The Tech Portal: Google rolls outs Gemma 3, its latest collection of lightweight AI models
• eWEEK: Google’s Gemma 3: Does the ‘World’s Best Single-Accelerator Model’ Outperform DeepSeek-V3?
• The Tech Basic: Gemma 3 by Google: Multilingual AI with Image and Video Analysis
• Analytics Vidhya: Google’s Gemma 3: Features, Benchmarks, Performance and Implementation
• www.infoworld.com: Google unveils Gemma 3 multi-modal AI models
• www.zdnet.com: Google claims Gemma 3 reaches 98% of DeepSeek's accuracy - using only one GPU
• AIwire: Google unveiled open source Gemma 3, is multimodal, comes in four sizes and can now handle more information and instructions thanks to a larger context window. The post appeared first on .
• Ars OpenForum: Google’s new Gemma 3 AI model is optimized to run on a single GPU
• THE DECODER: Google DeepMind has unveiled Gemma 3, a new generation of open AI models designed to deliver high performance with a relatively small footprint, making them suitable for running on individual GPUs or TPUs.
• Gradient Flow: Gemma 3: What You Need To Know
• Interconnects: Gemma 3, OLMo 2 32B, and the growing potential of open-source AI
• OODAloop: Gemma 3, Google's newest lightweight, open-source AI model, is designed for multimodal tasks and efficient deployment on various devices.
• NVIDIA Technical Blog: Google has released lightweight, multimodal, multilingual models called Gemma 3. The models are designed to run efficiently on phones and laptops.
• LessWrong: Google DeepMind has unveiled Gemma 3, a new generation of open AI models designed to deliver high performance with a relatively small footprint, making them suitable for running on individual GPUs or TPUs.

Classification:

• HashTags: #Gemma3 #OpenSourceAI #GoogleAI
• Company: Google
• Target: AI Developers
• Product: Gemma 3
• Feature: Open Source AI Model
• Type: AI
• Severity: Informative