Modes, Models & Generative Magic | Weekly Edition

PLUS HOT AI Tools & Tutorials

Aug 01, 2025

Hey there! Welcome to your latest Creators’ AI Edition!

This week, ChatGPT steps into the classroom with Study Mode, Microsoft launches Copilot Mode in Edge, and Chinese startup Z.ai drops a massive new open model that’s giving top players a run for their tokens. FLUX.1 Krea is redefining photorealism in AI images, Claude is getting usage caps, and Google’s Search AI just got a serious upgrade. Meanwhile, Runway’s Aleph is rewriting how we edit video, and NotebookLM is turning research into narrated explainers.

As always, we’ve rounded up the sharpest tools, guides, memes, and must-reads—let’s dive in!

Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁

Featured Materials 🎟️

ChatGPT as a Teacher

OpenAI has introduced Study Mode for ChatGPT, a new feature aimed at helping students learn by guiding them through problems step-by-step using Socratic questions and feedback instead of simply giving answers. Study Mode was developed with input from teaching experts and focuses on interactive learning, asking guiding questions, offering hints, and checking knowledge to encourage deeper understanding and active participation. In this mode, ChatGPT will resist giving quick solutions, steering students back into the learning process. The feature is rolling out now for Free, Plus, Pro, and Team users, with educational institutions set to get access in the coming weeks.

As AI becomes a bigger presence in education, Study Mode could help make it more of a genuine learning partner than just a shortcut, but its impact will likely depend on whether students choose to engage with the process.

Microsoft launched “Copilot Mode” in Edge

Microsoft has introduced Copilot Mode in Edge, bringing its AI assistant right into the browser to search across open tabs, manage tasks, and even proactively suggest or take actions. Copilot Mode is now built into Edge’s new tab page and features tools like voice input and multi-tab analysis, making the browsing experience much smarter. It’s launching free for now on both Windows and Mac with opt-in activation, though Microsoft has hinted that subscription pricing is coming in the future. Eventually, Copilot will be able to access browser history and credentials (with user permission) to handle tasks like completing bookings or running errands.

With Edge joining the agentic browser wars alongside competitors like Perplexity’s Comet and TBC’s Dia, it’s clear that AI-driven browsing is quickly becoming the new norm, even as the technology continues to evolve.

Z AI New Chinese Models

Chinese AI startup Z.ai (formerly Zhipu) has launched GLM-4.5, a new open-source agentic model family that offers aggressive pricing and delivers performance close to today’s top models in reasoning, coding, and autonomous tasks. GLM-4.5 comes with 355 billion parameters and is built to handle reasoning, code generation, and agentic workflows using a hybrid approach that balances speed and complexity.

According to Z.ai, GLM-4.5 now leads all open-source models worldwide and is just behind heavyweights like o3 and Grok 4 in overall capability. On agentic task and tool-use benchmarks, GLM-4.5 surpasses other open models, including o3 and Gemini 2.5 Pro, and even outperforms Grok 4with a 90% success rate in tool use.

Both GLM-4.5 and the lighter 4.5-Air version are being released with open weights, and Z.ai is also open-sourcing its innovative ‘slime’ training framework. With rapid innovation from Chinese labs like Qwen, Kimi, DeepSeek, MiniMax, and now Z.ai, the pace of open model development is accelerating and putting pressure on global AI leaders to keep up.

It looks better than Kimi and Qwen in our Ball test:

News of the week 🌍

ChatGPT Agent Available

ChatGPT Agent is now available to all Plus and Team subscribers, giving users access to AI assistants that can handle complex, multi-step tasks like planning, researching, shopping, and creating presentations. These agents use OpenAI’s most advanced models and have access to a “virtual computer,” which allows them to interact with web browsers, terminals, and a variety of online tools.

New FLUX.1 Krea

Black Forest Labs and creative platform Krea have unveiled FLUX.1 Krea, an open-weight image generation model designed to push photorealism to a new level and move beyond the telltale “AI look.” Drawing on a carefully curated and diverse training set, FLUX.1 Krea addresses common weaknesses seen in AI-generated images—such as waxy skin textures, blurry details, and unnatural color saturation—delivering strikingly lifelike visuals. The companies highlight that FLUX.1 Krea matches or surpasses the state-of-the-art among open models, and holds its own against top closed-source competitors (including Black Forest Labs’ own FLUX 1.1 Pro) in blind preference testing. The new release is fully integrated with the FLUX.1 [dev] ecosystem, making it accessible for developers and easy to implement across a variety of platforms.

The debut of such a high-quality open model signals a major leap forward in solving the so-called “AI aesthetic” problem, and as the line between real and AI-generated images blurs even further, it’s clear the next era of visual content is arriving fast.

Limits for Claude

Anthropic is introducing new weekly rate limits for Claude users starting August 28, aiming to reduce abuse and maintain service stability. These caps are mainly aimed at a small group of heavy users, especially those running Claude Code nonstop or reselling access. While the current 5-hour usage windows stay in place, the new weekly limits—including a specific cap for Claude Opus 4—will impact Pro and Max subscribers. Anthropic says less than 5% of users will be affected, though details on exact allowances haven’t been shared. This move comes as AI providers face growing demand and ongoing compute shortages.

Updates in Google Search AI

Google Search is rolling out new AI Mode, powered by Gemini models. You can now upload images and PDFs to get summaries or deeper info, use video input for real-time voice conversations with Search, and access AI Overviews through Lens in Chrome. Soon, Canvas in AI Mode will help you build and organize study plans with your files and web info, all in one place. To try these updates, opt in to the AI Mode Labs experiment here.

Aleph by Runway

Runway has introduced Aleph, a next-generation video model that allows users to edit and transform footage just by typing text prompts—no reshoots required. With Aleph, you can create new camera angles, change lighting, add or remove objects, and even extend shots in sequence, giving creators and filmmakers much greater control in post-production. Early access is available now, with a wider release coming soon, making Aleph a major leap forward for AI-driven video editing and creative workflows.

New Features in NotebookLM

NotebookLM, Google’s AI-powered research assistant, just introduced Video Overviews — narrated slide presentations that visually explain complex ideas using images, diagrams, and quotes from your documents. This makes the AI assistant far more helpful for learning abstract concepts, especially when visuals matter. Users can tailor each Video Overview based on expertise level or topic focus.

Alongside this, the Studio panel now lets users create and store multiple outputs (like Audio Overviews, Mind Maps, or Reports) per notebook — a major boost for collaboration, localization, and deep study. The upgraded Studio also supports multitasking, so you can listen to an Audio Overview while working on other outputs.

Useful tools ⚒️

CopyCat - Build browser automations with AI

Nitrode - AI game engine to prototype 3D games in a day

Kombai - The first AI agent built for real-world frontend tasks

Singify AI Vocal Remover - Remove vocals from any song

Rustic AI - Your visual AI design editor

Rustic AI is a next-generation design editor powered by AI. Generate visuals with prompts, then design freely using drag-and-drop tools — from product photos to social content. All in one place, no design skills required.

Share this post with friends, especially those interested in AI!