Welcome back!

OpenAI is reshuffling their roadmap. After cancelling o3-full release earlier this year, the company now intends to launch both o3-full and o4-mini soon, while delaying GPT-5 back to give it more time to grow stronger. Let’s unpack…

Today’s Summary:

• 🚀 OpenAI changes GPT-5 release plans

• 🔥 Google unveils massive GenAI upgrades

• 🧠 ChatGPT memory gets smarter upgrade

• 🖼️ HiDream open-source image model

• 💸 Anthropic debuts Claude Max plan

• 🔧 xAI launches Grok 3 API

• 🛠️ 2 new tools

TOP STORY

OpenAI changes plans: o3-full & o4-mini coming before GPT-5

The Summary: OpenAI CEO Sam Altman announced a surprise change to the release schedule. After previously canceling the launch of o3-full, OpenAI now plans to release both o3-full and o4-mini soon. GPT-5 has been delayed by "a few months" but is said to be much better than initially planned. OpenAI cited integration challenges and the need to prepare for "unprecedented demand".

Key details:

• OpenAI plans to release o3, o3 Pro, and o4-mini models very soon

• GPT-5 is delayed to allow for major upgrades and to handle future demand

• Free ChatGPT users will eventually get unlimited access to GPT-5 at the "standard intelligence setting"

• Mentions of “o4-mini” and “o3” already appear in ChatGPT web app code, hinting at an imminent launch

Why it matters: This pivot could point to integration challenges inside OpenAI, but also to major improvements expected for GPT-5. The decision to release multiple models may indicate OpenAI is responding to market pressure while buying time to polish its next big leap.

GOOGLE DEEPMIND

Google drops massive AI upgrades across API & research

The Summary: At Cloud Next 25, Google rolled out a range of generative AI upgrades spanning music, maps, voice, video, image, and hardware. The updates reflect Google’s strategy to embed generative AI into its APIs and platforms, making it accessible across enterprise and developer tools.

Key details:

• Lyria generates music from text prompts

• Veo 2 video generator adds editing and camera movement

• Chirp 3 clones voices from just 10 seconds of audio

• Gemini powers geospatial reasoning over maps

• Imagen 3 adds image editing with object removal and inpainting

• Ironwood is Google’s 7th-gen TPU high performance AI chip

• Agent Development Kit (ADK) is an open-source framework for building modular, multi-agent systems

Why it matters: Google is scaling generative AI across a range of interfaces, models, hardware, and real-world use cases. It’s building a tightly integrated platform where generative tools power everything from creative apps to enterprise systems.

FROM OUR PARTNERS

Today’s partner

Seeking impartial news? Meet 1440.

Every day, 3.5 million readers turn to 1440 for their factual news. We sift through 100+ sources to bring you a complete summary of politics, global events, business, and culture, all in a brief 5-minute email. Enjoy an impartial news experience.

Join for free today!

OPENAI

ChatGPT gets memory upgrade

The Summary: OpenAI is rolling out an upgrade to ChatGPT’s memory, letting it reference the full conversation history across text, voice, and image prompts. This means it can now remember context from previous chats automatically. The feature rolls out first to Plus and Pro users, with Team and Enterprise support coming soon.

Key details:

• ChatGPT can now use all past chats as memory

• New toggle in settings separates “chat history” and “saved memories”, and each type of memory can be turned off

• Temporary Chat disables memory like an incognito mode

• Sensitive info like health data won’t be remembered unless you explicitly request it

• Rollout excludes EU, UK, Switzerland, Norway, Iceland, and Liechtenstein due to regulatory review

Why it matters: Pulling information from your entire chat history means ChatGPT can personalize responses based on your preferences and recurring topics. Each new chat can build on everything you’ve shared before. This changes the relationship between you and ChatGPT, knowing that anything you say might echo back later.

How-to guide

QUICK NEWS

Quick news

• xAI launches an API for Grok 3

• Anthropic rolls out Claude Max high limits plan

• HiDream-I1 new open-source state of the art image model

TOOLS

🥇 New tools

• Firebase Studio - Free AI dev tool to build full-stack apps

• Voicenotes Pages - Audio-first publishing tool to share voice notes