GPT-5 Is an Evolution, Not a Revolution
OpenAI stole the show this week, but Google and Anthropic weren’t quiet either.
Welcome to The Median, DataCamp’s newsletter for August 8, 2025.
In this edition: OpenAI launched GPT-5 and two open-weight models, Google announced Genie 3 and released Deep Think, while Anthropic updated Claude Opus.
This Week in 60 Seconds
OpenAI Launches GPT-5, Unifying Previous Models
OpenAI has released GPT-5, its new flagship model that replaces and unifies the previous generation of models like GPT-4o and o3. The new system automatically routes user prompts to decide in real-time whether to provide a quick answer or engage in slower, deeper reasoning. While the update is more of a user experience overhaul than a massive capability leap, it introduces new features like preset personalities, chat color customization, and, for paid users, integrations with Gmail and Google Calendar. The context window remains limited, with 8K tokens for free users, 32K for Plus, and 128K for Pro subscribers. We’ll explore GPT-5 in the “Deeper Look” section.
OpenAI Releases gpt-oss, Its First Open-Weight Model Since GPT-2
On August 5, OpenAI released gpt-oss-120b and gpt-oss-20b, two new open-weight language models licensed under the flexible Apache 2.0 license. The models are designed for strong performance on reasoning tasks at a low cost and are optimized for consumer hardware. The larger 120B model reportedly achieves near-parity with OpenAI's o4-mini on core reasoning benchmarks while being capable of running on a single 80GB GPU. The smaller 20B model is designed for edge devices and local inference, requiring only 16GB of memory. We’ve written a tutorial on how to run these models on your own laptop.
Anthropic Releases Claude Opus 4.1
Anthropic has released Claude Opus 4.1, an upgrade to its most advanced model. The new version shows improved performance on agentic tasks, coding, and reasoning, achieving a state-of-the-art score of 74.5% on the SWE-bench Verified coding evaluation. According to early partner feedback, the model excels at pinpointing corrections in large codebases without introducing new bugs. Opus 4.1 is now available to paid Claude users, via the API, and on cloud platforms including Amazon Bedrock and Google Cloud's Vertex AI at the same price as its predecessor.
Google DeepMind Announces Genie 3, a Real-Time Interactive World Generator
Google DeepMind has announced Genie 3, a "general-purpose world model" capable of generating interactive, dynamic environments from text prompts. The model allows users to navigate these generated worlds in real-time at 24 frames per second with a 720p resolution, maintaining consistency for several minutes. Unlike previous versions, Genie 3 is the first in the series to enable real-time interaction. Genie 3 is currently available as a limited research preview for a small group of academics and creators.
Google Rolls Out Gemini 2.5 Deep Think
Google has released Deep Think in the Gemini app for Google AI Ultra subscribers. This is not to be confused with deep research—this feature uses "parallel thinking techniques," allowing the model to explore multiple lines of reasoning at once to solve complex problems. This release is a faster, more user-friendly version of the model that recently achieved a gold-medal standard at the International Mathematical Olympiad (IMO). While not the full IMO version, this model achieves Bronze-level performance on the same benchmarks and excels at creative tasks like iterative design and tough coding challenges.
New Course: Introduction to SQL With AI
A Deeper Look at This Week’s News
What’s New in GPT-5?
GPT-5 isn’t the AGI milestone some were hoping for, and it certainly doesn’t feel like having a “PhD in your pocket.” However, it’s a well-executed consolidation of OpenAI’s previous lineup into a unified experience.
Overview
If you’ve been used to seeing options like GPT-4o, GPT-4o-mini, or o3 in the model picker, those are gone. You no longer have to decide which one to use for speed or quality—the system now does that automatically.
When you type a prompt, GPT-5’s router decides in real time whether to give you a fast response or engage in deeper, slower reasoning. The goal is to make the experience seamless: one model name, consistent behavior, and no manual switching.
As you can see in the image above, you can still manually select GPT-5 Thinking if you want the model to take extra time and deliver more thorough, step-by-step answers, or GPT-5 Pro if you need maximum reasoning depth and accuracy for research-grade tasks. The difference is that these are now variations of the same core model.
Here’s how the new family lines up against the previous generation:
Source: OpenAI
What you get depending on your tier
The free version gives you access to the main GPT-5 model as well as GPT-5 Thinking, but with the smallest context window and tighter usage limits. It’s fine for everyday chatting, short drafting, or answering questions, but you’ll quickly hit the ceiling if you try to work with longer documents.
Plus subscribers get the same models but with expanded usage and a bigger 32K token context window—enough to handle medium-sized PDFs or a more sustained back-and-forth before things fall out of memory. Response times are also noticeably faster here compared to the free tier, which is throttled based on availability.
Pro is where things open up. You get GPT-5, GPT-5 Thinking, and GPT-5 Pro—the high-end variant designed for maximum reasoning depth and accuracy. The context window jumps to 128K tokens, which is large enough for book-chapter-level work or multiple long files in a single session.
Team and Enterprise plans are essentially custom arrangements, but they include all variants, flexible usage, and the fastest response times available. Enterprise users also get the 128K context window, while Team sticks to 32K.
New features in GPT-5
Customize the color of your chats
You can now pick the color scheme for your chats. It’s purely cosmetic, but it helps make the interface more like your own environment. You can change the color from the General section in Settings:
Change personalities
GPT-5 introduces preset personalities, allowing you to switch the assistant’s style to be more supportive, concise and professional, or even lightly sarcastic. Because GPT-5’s steerability has improved, these styles stick throughout the conversation instead of fading after a few replies.
To access this feature, go to the Personalization section in Settings, click Custom instructions, and then select the personality you want by picking a preset:
Gmail and Google Calendar integration
For Plus, Pro, Team, and Enterprise users, GPT-5 can connect directly to your Gmail and Google Calendar. It can pull in your schedule, help you find free time, and even draft responses to emails you’ve been ignoring. It’s a genuine step toward having an AI actively manage your day.
To use this feature, go to the Connectors section in Settings and follow the on-screen instructions to connect your Gmail and Google Calendar.
Safer, more useful completions
GPT-5 replaces the old refusal-based safety approach with “safe completions.” Instead of simply blocking a request that might be unsafe, it gives you as much helpful, safe information as possible while explaining any limitations. It also reduces sycophancy.
To read more about GPT-5, read our full blog post here.
Industry Use Cases
AI Strengthens Infrastructure Against Climate Disasters
According to Deloitte, strategically applying AI across the infrastructure lifecycle—from planning to disaster response—could prevent up to 15% of projected annual damage by 2050. Use cases include AI-powered digital twins that simulate future climate conditions and early warning systems that detect bushfires and floods with greater accuracy. Lisbon is already seeing success: AI-enhanced models helped design a drainage system to mitigate 20 future floods. In the recovery phase, AI pinpoints damage hotspots and accelerates resource deployment, helping communities rebuild faster and smarter.
AI Simplifies Machine Programming in Manufacturing
In an industry long reliant on specialized expertise, AI is making computer-aided manufacturing (CAM) more accessible and efficient. CAM software translates digital part designs into instructions for machines that cut or shape materials, but it’s often complex and difficult for newcomers to use. To address this, Mastercam has added AI-powered Help and Command tools that guide users through toolpath creation with fewer clicks and less trial-and-error. The update helps teams adapt to labor shortages, shrinking production runs, and tighter supply chains by accelerating setup and freeing experts to focus on more advanced tasks.
AI Rethinks Critical Mineral Discovery
Startups like Earth AI and Terra AI are using machine learning to radically speed up mineral exploration. Earth AI claims a 75% success rate—far above industry norms—by training its predictive algorithms on decades of geological data, drilling its own test sites, and verifying finds such as indium and palladium in Australia. Terra AI, meanwhile, uses AI to process magnetic and seismic data, generating thousands of underground maps to target high-potential drill zones. Legacy firms like Exiger are also tapping into AI to map vulnerabilities in mineral supply chains, helping clients reduce geopolitical dependencies. Though adoption faces regulatory hurdles, industry leaders say AI could halve the time to bring new mines online.
Tokens of Wisdom
The power of private data is becoming more important. If everyone is using the same pre-trained model, then the only way we make differences among ourselves is by the data we have.
—Jure Leskovec, Pioneer of Graph Transformers, Professor at Stanford
Listen to our latest DataFramed podcast, where Richie and Jure explore foundation models for enterprise data, the limitations of current AI models in predictive tasks, the potential of graph transformers for business data, relational foundation models for ML workflows, and much more.