Welcome to Neural. AI evolves rapidly. We help you navigate through it. OpenAI has announced the upcoming release of GPT-4.5 for ChatGPT within weeks. Meanwhile, xAI is set to unveil Grok 3 later tonight. Let the countdown begin for OpenAI to execute its most entertaining move yet…
Grok 3 Demo Day
Hot on the heels of hinting at a potential acquisition of OpenAI, Elon Musk confirmed on Saturday that Grok 3 will debut tonight via a live demonstration set for 8 p.m. PT.
Musk has been promoting Grok 3 as the “smartest AI on Earth.” Exact details regarding Grok 3 will remain under wraps until tonight’s announcement.
Here’s a brief timeline of xAI and Grok’s journey:
- xAI founded in July 2023
- Grok-0 completed in August 2023
- Grok-1 previewed in November 2023
- Grok-1.5 finished in March 2024
- Grok-2 launched in August 2024
As of 2025, xAI has transitioned Grok from a feature in the X app to a full-fledged app and website.
One of my pressing queries about Grok is: When can we expect it in Tesla vehicles? While AI-assisted (supervised) driving is invaluable, a Grok-integrated voice assistant could enhance user interaction beyond the existing voice controls.
Another question is whether OpenAI will overshadow xAI’s showcase tonight by introducing something exciting with GPT-4.5.
This Week in OpenAI
Last week, OpenAI’s Sam Altman outlined the future of ChatGPT. He mentioned that GPT-4.5 would be available in a few weeks, hinting that a launch today is improbable, with GPT-5 expected in the following months.
GPT-5 aims to merge fast large language models with reasoning models that take longer to process—eliminating the need to sift through multiple models per query.
In the meantime, Altman continues to create buzz around GPT-4.5 on X:
When asked if he intended to “steal the show” tonight, Altman responded tactfully, saying, “that wouldn’t be very nice…”
Separately, OpenAI has rolled out a few tangible releases in recent days:
- OpenAI “o1 and o3-mini now allow both file and image uploads in ChatGPT”
- OpenAI “increased o3-mini-high limits for Plus users by 7x, now up to 50 per day”
- OpenAI published a Model Spec update outlining desired behaviors for its AI models
An impressive yet nebulous hype post was also shared:
This update aims to enhance GPT-4o’s writing capabilities, especially when given examples, while reducing its tendencies toward generative sloppiness.
Recent developments at OpenAI highlight the increasing importance of GPT-5 for ChatGPT. The complexity of identifying which models excel at specific tasks is becoming cumbersome and often outdated before implementation.
Anthropic Isn’t Overlooking Claude
Among the noteworthy updates from recent days is the impending release of Claude 4 by Anthropic. Similar to GPT-5, Claude 4 is anticipated to be a unified chatbot, capable of providing quick responses like Claude 3.5 while also tackling more intensive reasoning queries.
Claude 3 debuted last March, followed by Claude 3.5 in June, so the next significant upgrade is on the horizon. Keep an eye out for an official announcement.
More
In other AI news, OpenAI’s board formally declined Elon Musk’s $97.4 billion acquisition offer; Mira Murati’s startup has onboarded another OpenAI co-founder; France’s Mistral has launched a regional model focused on Arabic language and culture; AI search platform Perplexity has released Deep Research, which is distinctly unconnected to OpenAI’s similarly-named product; and Thompson Reuters triumphed in the first significant AI copyright case in the U.S., ironically against the legal AI firm Ross Intelligence.
Bringing this full circle, Grok 3 has been trained on “all court cases,” which we can expect to see showcased tonight.
Stay tuned for more on the latest AI advancements in the next installment of Neural— exclusively on DMN! View the previous edition here.
