Anapoly AI Labs

Tag: chatgpt-5

  • ChatGPT-5 Availability and Features

    Ray passed on this useful summary of the new features in ChatGPT-5 and their availability to the various tiers of user.

  • ChatGPT-5 Availability and Features

    ChatGPT-5 Availability and Features

    in August 2025, as described by Gemini AI

    Transparency label: AI-only

    Based on the latest information, ChatGPT-5 has been released and is now available to all users, though with varying access levels and features depending on the subscription plan.

    Here’s a breakdown of the availability and features for paying subscribers:

    Availability for All Users:

    • Default Model: ChatGPT-5 is now the default model for all users, replacing previous versions like GPT-4o.
    • Unified System: The new system is designed to automatically select the best model for a given task. It has a fast, general-purpose model for everyday queries and a more advanced “GPT-5 Thinking” model for complex reasoning tasks. A built-in router decides which one to use.
    • Usage Limits: Free users have a cap on the number of messages they can send. Once this limit is reached, the system automatically switches to a smaller, “mini” version of the model to maintain functionality.

    Additional Features for Paying Subscribers (Plus, Pro, and Team Plans):

    • Increased Usage Limits: Paying subscribers have significantly higher message caps, allowing for more extensive use of the powerful models.
    • Model Picker: Paid users have access to a model picker, allowing them to manually switch between the standard GPT-5 model and the “GPT-5 Thinking” model for more control over their responses.
    • GPT-5 Pro (Pro and Team Tiers): Subscribers on the Pro and Team tiers get access to a specialized version, “GPT-5 Thinking Pro.” This model takes a bit longer to respond but provides more accurate results for complex and demanding tasks.
    • Advanced Capabilities: In addition to the core model improvements, paying subscribers also get access to a wider range of features, including:
      • Advanced Voice Mode: Enhanced, natural voice conversations.
      • Image Generation: Higher usage limits for the DALL-E 3 image generation tool.
      • Custom GPTs: The ability to create and use custom, specialized versions of the chatbot.
      • File Analysis: The ability to upload and analyze documents and images.
      • Deep Research Tools: Access to specialized tools for fields like finance, science, and engineering.
      • Priority Access: Fewer interruptions during high-traffic periods.

    New Features in ChatGPT-5 for All Users:

    The new GPT-5 model itself brings significant improvements across the board, including:

    • Improved Reasoning: The model is now capable of more complex and accurate reasoning.
    • Reduced Hallucinations: It’s more factual and reliable, with a significant reduction in generated false information.
    • Enhanced Writing and Coding: The model produces higher-quality, more coherent writing and more functional, aesthetically-pleasing code.
    • Better Instruction Following: It is much better at understanding and executing complex, multi-step instructions.
    • Advanced Health Capabilities: GPT-5 has an improved ability to understand and interpret medical terminology, and can even flag potential health risks based on user input, though it’s not a substitute for professional medical advice.

  • Creating a video from reference documents

    Transparency label: this post is Human-only; the video is AI-only

    In an earlier post, I explained how I used NotebookLM to update our Contextual Scaffolding Framework to allow for new capabilities in ChatGPT-5. In that same session, I accidentally clicked on the “Video overview” function, and this is what NotebookLM produced. I did not give it any instructions about focus or audience, and it was working only with the then current version of my Contextual Scaffolding Framework and the ChatGPT-5 Prompting Guide. I think it’s a remarkably good. See what you think.

  • GPT-5, the Router, and the Road to a SuperApp

    Transparency label: AI-only

    OpenAI’s latest release, GPT-5, isn’t just about new features or smarter answers, it’s a strategic move to turn ChatGPT’s 700 million free users into a sustainable business. The real engine behind this shift is the new “router” system, which decides in real time which AI model to use for each request.

    SemiAnalysis has an insightful breakdown of how this technology could underpin a new monetisation model: ChatGPT as a “SuperApp” that can act as your purchasing agent, make bookings, and complete transactions—all without sending you to a search engine.

    Read a summary of the key points here.

  • Monetising ChatGPT-5

    Monetising ChatGPT-5

    Transparency label: AI-only

    The newsletter “GPT-5 Set the Stage for Ad Monetization and the SuperApp“ from SemiAnalysis outlines OpenAI’s strategic pivot towards monetising its vast free user base, a move largely enabled by the GPT-5 release and its new “router” system.

    GPT-5, while underwhelming for many “power users,” is designed with a different audience in mind: the 700 million-plus free ChatGPT users who use the service infrequently and remain unmonetised. This group represents OpenAI’s largest untapped opportunity.

    At the heart of GPT-5 is the “One United System”—a real-time router that decides which model to deploy based on the conversation’s type, complexity, required tools, and user intent. By learning continuously from user signals such as model switching, preference rates, and measured correctness, the router improves over time. It enables lower-cost usage through mini-versions of models while, for the first time, giving many free users access to advanced “thinking” or Chain of Thought reasoning.

    This router is more than a technical upgrade; it is the foundation for monetising free users. Its ability to understand query intent paves the way for assigning commercial value to interactions. That vision is reinforced by strategic hires, notably Fidji Simo as CEO of Applications, whose experience at Facebook and Instacart includes successfully monetising high-intent internet properties through ad products and agent-based checkout. Sam Altman’s own tone has shifted from dismissing ads to acknowledging the potential for transaction-based or affiliate revenue in ways that could avoid intrusive advertising.

    A central part of the monetisation strategy is what SemiAnalysis calls an “agentic SuperApp.” This would see ChatGPT acting as an active purchasing agent, handling not just recommendations but full-service transactions in areas like groceries, e-commerce, flights, and hotels. Unlike traditional search advertising, large language models have marginal costs: more computation can yield more valuable results. The router could allocate substantial resources to high-value queries, such as researching and contacting top local lawyers, because the potential referral revenue could run into thousands of dollars. In this model, revenue would come from transaction fees or ad take rates on purchases, not subscriptions.

    Elements of this future are already in development. ChatGPT is integrating with Gmail and Google Calendar, benchmarking tool use in industries such as telecoms and retail, and building towards partnerships with platforms like Stripe, Shopify, Booking.com, and Instacart. Such collaborations could offer companies lower customer acquisition costs, making the model attractive to partners.

    If successful, OpenAI could evolve ChatGPT into a consumer SuperApp, competing directly with Google, Meta, and Amazon by bypassing traditional search funnels and ad-tech ecosystems—an ambitious path that, if executed, could reshape both AI usage and online commerce.