Google’s latest AI fits in your pocket

Welcome back.

Google’s new Gemma 3 just dropped, and it’s small enough to run smoothly on your phone but powerful enough to outperform top-tier models.

Is this the everyday AI we’ve been waiting for?

Let’s find out.

In today’s release:

1. Google launches a new AI model

2. Generate images with Gemini 2.0 Flash

3. Manus AI is teaming up with Alibaba

Google launched Gemma 3

Google has unveiled Gemma 3, its latest generation of open-source AI models. The model is designed to run efficiently even on modest hardware, from phones to laptops.

Gemma 3 comes in four sizes: 1B, 4B, 12B, and 27B parameters, and outpaces rivals including Llama3-405B, DeepSeek-V3, and o3-mini, according to initial benchmarks.

• Performance: Delivers top-tier results in complex tasks with significantly lower hardware requirements.

• Multilingual: Supports 140 languages, ideal for global applications.

• Large context: Processes inputs of up to 128k tokens (about a 300-page book).

• On-device capabilities: Runs efficiently on devices from smartphones to single GPUs, allowing for flexible integration.

Google also introduced ShieldGemma 2, a safety-focused companion model designed to automatically flag harmful or unsafe images.

How can you use this to your advantage?

Gemma 3 enables you to incorporate advanced AI into products without hefty computing resources, streamlining deployment on everyday devices.

Generate AI images with Gemini 2.0 Flash

Google’s Gemini 2.0 Flash can natively generate images from text prompts. Unlike previous models that combined language models with separate image generators, Gemini 2.0 Flash integrates these capabilities, allowing for more accurate and versatile image creation.

You can experiment with this feature using the experimental version, gemini-2.0-flash-exp, available in Google AI Studio and via the Gemini API.

• Text and image storytelling: Generate illustrated stories with consistent characters and settings.

• Conversational image editing: Iteratively refine images through natural language prompts, enabling real-time collaboration.

• World knowledge-based generation: Leverage broader reasoning to produce contextually relevant images, such as detailed recipe illustrations.

• Improved text rendering: Accurately generate legible text within images.

Some early users have been impressed with its capabilities to edit photos using only text prompts and generate images in specific art styles.

How can you use this to your advantage?

Gemini 2.0 Flash enables businesses to incorporate advanced AI image generation directly into applications, enhancing content creation, marketing materials, and user engagement with more dynamic visuals.

Manus AI partners with Alibaba

Chinese AI startup Manus AI has announced a strategic partnership with Alibaba's Qwen team to enhance its autonomous AI agent capabilities.

This collaboration aims to integrate Manus' functionalities with Qwen's open-source AI models, optimizing performance on China's domestic computing platforms.

• Enhanced autonomy: Manus AI can independently execute complex tasks such as resume screening, stock analysis, and website development.

• Benchmark performance: Reports indicate that Manus AI has outperformed OpenAI's DeepResearch in GAIA tests, achieving state-of-the-art scores in real-world problem-solving.

This partnership underscores China's rapid advancements in AI technology and its potential implications for global business practices.

How can you use this to your advantage?

For businesses and developers, the integration of Manus AI with Alibaba's Qwen models offers access to advanced, autonomous AI capabilities. This can streamline operations, enhance data analysis, and improve decision-making processes, providing a competitive edge in various industries.

OTHER AI NEWS

Google launches Gemini Robotics: Gemini Robotics is an advanced AI model designed to help robots perform complex, real-world tasks. The technology could soon enable robots to safely understand, interact with, and adapt to dynamic environments, making robots smarter and safer for everyday use.

Microsoft introduces Xbox Copilot: Copilot for Gaming is an AI assistant for Xbox that provides personalized gameplay tips, game recaps, and coaching. Initially available in preview to Xbox Insiders in April, it aims to enhance the gaming experience through tailored, real-time advice.

Adobe Stock adds instant AI editing: Adobe Stock's new "Customize" feature lets you edit and refine stock images using AI, without Photoshop. Users can now quickly modify aspect ratios, styles, and image variations directly within Adobe Stock, saving time and effort.

Gemini app gets an upgrade: Gemini now personalizes responses, enhances research, and integrates with Google apps. Users can try Deep Research for free, create custom AI “Gems,” and access smarter, faster AI tools.

OpenAI pushes for ‘fair use’: OpenAI is urging the US to protect AI’s right to train on copyrighted data. It argues that limiting AI training would hurt innovation, despite ongoing lawsuits from content owners.

POPULAR AI TOOLS

1. Zencoder → Integrated, customizable, and intuitive coding agent.

2. Bolt x Figma → Turn Figma designs into production-ready apps in one click.

3. STORI → AI building your inbound.

4. Eddie AI 2.0 → The assistant editor for people who actually create videos.

5. Waitlist.Email → Unlimited, unbranded email waitlists for developers.

AND THAT’S A WRAP

Thank you for reading!

If you found this email useful, share it with a friend or colleague who also loves AI.

Also, drop me a follow on Twitter/X for more AI and tech updates.

I will talk to you soon!

Mike