OpenAI’s X and new AI models: The start-up’s latest big moves

In a move that would deepen the hostility between Musk and Altman, OpenAI could be building its own X-like platform.

Since going mainstream, OpenAI has made constant headlines for its product launches and technological developments, and this past week has been no different.

Over the last few days, the $300bn company has launched new reasoning models and is reportedly working on its own social media platform, among other things.

This, alongside releasing its latest multimodal AI model, the GPT-4.1, just days ago, which the company claims features “major improvements”. Here is a rundown of the latest OpenAI has been up to.

An OpenAI social media platform?

On 15 April, The Verge reported that OpenAI is working on building an X-like social media platform. Sources told the tech publication that there’s an “internal prototype” focused on ChatGPT’s image generation with a social feed.

The start-up’s CEO Sam Altman has been reaching out to outsiders for feedback, the report read.

Altman and X-owner Elon Musk have been at odds for the past few years. Musk, who is an OpenAI co-founder, sued the company last year for allegedly not sticking to its original objective of developing AI for the benefit of humanity.

More recently, Musk, who is also a close aide to US president Donald Trump, made an unsolicited $97.4bn bid to purchase the non-profit that owns OpenAI.

However, he later claimed he would withdraw the offer if the AI start-up’s board of directors agree to stop the company’s for-profit transition.

Two new reasoning models

On Wednesday (16 April), OpenAI released its latest reasoning models, the o3, which OpenAI claims is its “most powerful reasoning model” and o4-mini, which is a small model optimised for “cost-efficient reasoning”.

According to the company, o3 “pushes the frontier” across a range of activities, especially at visual tasks such as analysing images, charts and graphics.

While o4-mini outperforms o3-mini, its predecessor, on non-STEM tasks, as well as in areas such as data science, the company claimed. Moreover, OpenAI said that the new small reasoning model supports “significantly higher usage limits” when compared to o3.

“Compared to previous iterations of our reasoning models, these two models should also feel more natural and conversational,” OpenAI said.

Alongside releasing its latest reasoning models, the AI company also launched Flex processing, an API option that offers slower response time and “occasional resource unavailability” for lower costs.

According to the company, the offering is “ideal for non-production or lower-priority tasks such as model evaluations, data enrichment or asynchronous workloads”.

Flex processing is currently only available in its o3 and o4-mini models in a beta version, it said.

A possible new acquisition 

Windsurf, the AI coding assistant, is in talks to be acquired by OpenAI for around $3bn, Bloomberg reported earlier this week. Last August, the company, formerly known as Codeium, raised $150m in a Series C funding round at a valuation of $1.25bn. And earlier this year, it was in talks to raise funding at a $3bn valuation.

With this acquisition, OpenAI would be in direct competition with other AI coding assistant providers.

While yesterday, a CNBC report broke that last year the AI behemoth made an approach about acquiring Anysphere, a start-up the company is an investor in. But the talks failed to proceed.

Instead, Anysphere has been in talks with venture capitalists to raise funds at a $10bn valuation, Bloomberg reported early last month.

If the Windsurf acquisition goes through, the OpenAI-owned business will be competing with Anysphere.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.