Manus AI Now Open to All🤩

AI continues to push boundaries across industries, with three major developments redefining what’s possible this week. Manus AI, a powerful autonomous coding agent, is now publicly available without a waitlist, offering users 1,000 free credits plus 300 daily credits to experience hands-free software development. This allows developers to delegate entire projects to an AI that plans, codes, and executes tasks across multiple programming languages and frameworks—accelerating workflows with minimal effort. In healthcare, OpenAI has introduced HealthBench, a new benchmark co-created with 262 physicians across 60 countries to evaluate AI performance in real-world clinical scenarios. Featuring over 5,000 multilingual conversations and more than 48,000 unique evaluation criteria, HealthBench aims to ensure AI systems are safe, accurate, and aligned with physician judgment. At the intersection of medicine and computer vision, researchers have also developed FaceAge, a deep learning model that estimates biological age from facial images. Tested across thousands of cancer patients, the model was found to predict survival outcomes more accurately than traditional measures and shows promise as a biomarker for aging. Together, these advancements signal a future where AI not only builds our digital tools but also plays an integral role in safeguarding and understanding human health.

Manus AI: Autonomous AI Agent Now Live

After waiting so long, Manus AI has finally removed its waitlist and is now available to all users immediately, offering a one-time bonus of 1,000 free credits plus 300 daily credits for autonomous coding and development tasks. This major update gives users full access to Manus’s powerful AI agent, which independently plans, executes, and completes complex projects across multiple programming languages and frameworks. The move follows significant funding and growing competition in the AI space, making Manus a highly accessible and valuable tool for developers seeking advanced AI assistance without financial barriers. With these free credits and no waitlist, users can now experience Manus AI’s autonomous capabilities firsthand and accelerate their coding workflows with more freedom and flexibility than ever before.

OpenAI’s A Real-World Benchmark for Evaluating AI in Healthcare

HealthBench is a new benchmark developed to evaluate how well AI systems perform in realistic health scenarios, aiming to ensure that AI models are both useful and safe for healthcare applications. Created in partnership with 262 physicians from 60 countries, HealthBench consists of 5,000 multi-turn, multilingual health conversations that mirror real-life interactions between users or clinicians and AI, covering a wide range of medical specialties and complexities. Each conversation is graded using a detailed physician-created rubric, with over 48,000 unique criteria that assess whether the AI’s responses meet expert standards for accuracy, relevance, and safety. Model responses are scored by an AI grader (GPT-4.1), and the benchmark is designed to reflect real-world impact, align with physician judgment, and leave room for future model improvements. This approach supports the responsible development of AI in health by providing meaningful, trustworthy, and rigorous evaluation that encourages continuous progress and prioritizes patient safety.

Detecting Cancer by Reading Your Face

FaceAge is a deep learning system developed to estimate biological age from simple face photographs, aiming to provide a more objective and clinically useful measure of physiological health than chronological age. Trained on large datasets of healthy individuals and validated in over 6,000 cancer patients across multiple institutions, FaceAge was shown to predict survival outcomes independently of other clinical factors, with patients who "looked older" experiencing worse overall survival. On average, cancer patients appeared about five years older than their actual age compared to healthy controls, and integrating FaceAge into existing clinical prediction models improved the accuracy of survival forecasts, especially for those receiving palliative care. The study also found that FaceAge estimates were linked to molecular markers of aging, suggesting its potential as a biomarker. These results highlight that a person’s facial features, as analyzed by AI, can reflect biological aging and help doctors make better-informed treatment decisions, though further validation in larger and more diverse patient groups is needed.

Hand Picked Video

In this video, we’ll look at OpenManus, the ultimate open-source AI agent that lets you build and automate without restrictions, powered by GPT-4o and designed for seamless AI-driven workflows, from website creation to stock analysis, all completely free and accessible to everyone.

Top AI Products from this week

• Shadow - Stop recapping and start delivering. Shadow captures every insight, transforms decisions into action, and ensures you make the most of your meeting data. Turn discussions into real progress—with Shadow, all your conversations lead to results, not just notes.

• Gadget - Gadget is an AI-centered IDE with all of the infrastructure, tools, integrations, and frameworks you need under one roof. Code, configure, debug, deploy, and scale apps with an AI assistant that can do it all quickly, and ensure it's always done correctly.

• Dolphin AI - Dolphin finds and tracks feature requests, pain points, and customer love from your B2B calls - on autopilot. No more "What's the status of this feature?" or "Just add it to Productboard." Always keep up with what customers need and ask for.

• Generated Assets - Generated Assets is an interactive AI tool from Public.com that turns any idea into an investable index. Simply describe any investment idea in the prompt bar, and our AI will screen thousands of stocks to help you build a tailored index to invest in.

• Kazu Navi - Kazu Navi (かずナビ) is your guide to learning Japanese—starting with numbers and expanding into grammar, sentence structure, and context. Analyze tricky phrases, dive deeper with interactive modular tools. Learn with clarity, confidence, and cultural insight.

• Basin MCP - Basin MCP stops code gen hallucinations by extensively testing your copilot’s output and feeding test results back to the copilot so it can automatically improve its code. Tested w/ Cursor & Windsurf, works w/ any copilot that supports agentic MCP capabilities.

This week in AI

• OpenAI Open-Source Model - Sam Altman announced during Senate testimony that OpenAI plans to release an open-source AI model this summer, aiming to boost innovation and transparency.

• TikTok AI Alive Animates Photos - TikTok AI Alive lets users animate static photos into short videos with movement and effects, available via Story Camera. AI-generated content is labeled and safety-checked.

• Audible AI Narration & Translation - Audible partners with publishers to produce AI-narrated audiobooks via end-to-end or self-service options, offering 100+ voices in multiple languages. AI translation beta launches in 2025.

• Reasoning Model Scaling Limits - OpenAI’s o3 reasoning model scaled 10× from o1, but rapid growth may slow within a year as compute needs approach the frontier. Scaling may face data and research limits.