(More-so if you have any apps of your own, or if you are using some from elsewhere - and they suddenly stop working, then maybe reach out to check the creator knows about these deprecations.) It seems it's not just GPT's that are vanishing on Dec 11th. API access to the following six models will end on December 11, 2026: GPT-5 GPT-5 mini GPT-5 nano GPT-5 pro o3 o3-pro To avoid any disruption, they recommend migrating to the replacement models mentioned here https://developers.openai.com/api/docs/deprecations#2026-06-11-gpt-5-and-o3-model-deprecations before the shutdown date. If you have any questions about model migration, You can reach out on the OpenAI Developer Forum here: https://community.openai.com/c/api/deprecations/32 Hope that helps

As part of our continuous upgrade process, we are deprecating GPT-Image-2. Access to this model will shut off on February 10, 2027. Ahead of the shutdown timeline, we encourage you to explore GPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare, our state-of-the-art image generation models.

A Video from one of the guys at superhuman.ai that may be useful: https://www.youtube.com/watch?v=ZSqZ3y9WYZA Ever had AI create a graphic you like, then asked for something to match... and got a completely different look? This Claude tutorial shows a three-step approach to keeping your branding consistent. Using a coffee business as the example, it starts with one visual style guide and works through: 1. Create the designs: Business cards, a shop poster, and a matching website. 2. Check the results: Add a review step where Claude checks its work against the design requirements and flags problems. 3. Save the process: Package the workflow as a reusable Claude Skill for future projects. The useful lesson here is giving AI a clear set of design rules, then having it check whether it followed them. If you create graphics for your own products or clients, it’s worth watching how the pieces come together. Less explaining the same things again. More chance of getting designs that look like they came from the same business! Hope it helps

Its latest model for coding, cybersecurity, and complicated jobs that take more than a quick answer. https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/ Google says its teams have seen it work through engineering tasks for days, tackling problems and trying different approaches when it gets stuck. In internal testing, it also achieved more than twice the success rate of Gemini 4 Argon at generating secure code. For those of us building apps, better help finding bugs and fixing security weaknesses could be useful. But we’ll have to wait to try it. Access is starting with selected cybersecurity defenders, with a wider rollout planned. Introductory API pricing is $2 per million input tokens and $10 per million output tokens. That’s the usage pricing for connecting it to software, rather than a monthly chat subscription. The figures sound promising. What interests me is how well it handles an actual project when something awkward goes wrong! I guess we wait and see.

Last month, Open AI made a big move that basically closed down what was becoming the quickest and easiest way of creating a useful product. They made Custom GPTs non-shareable. Poo! But fine, everyone thought, at least all of the ones that were made before this happened could still be used by anyone with the link... Not all poo....Phew! Well, now it has been announced that all Custom GPTs will all be "sunsetted" in a couple of months time. Back to "poo" again!