Meta Claims Its New Model Has Caught GPT-5.5. The Benchmark Race Just Got Interesting.

Your next income stream could already be on this list.

Most people are still using AI to write emails. Others are using the same tools to build businesses, launch side hustles, and create entirely new ways to earn.

Get 200+ proven ways to make money with AI in 2026, plus the free daily AI newsletter trusted by 1M+ professionals.

1  Meta says its next flagship model has matched GPT-5.5 on key benchmarks.

Alexandr Wang, Meta's chief of superintelligence, reportedly told colleagues that the company's next model, internally codenamed Watermelon, has pulled level with OpenAI's GPT-5.5 on the metrics that matter most. Wang also hinted publicly at a near-term Muse Spark update targeting coding improvements aimed directly at Claude Opus. Meta is spending up to $145 billion on AI infrastructure this year, and benchmark parity with OpenAI would be a significant milestone for a company that has consistently trailed on model performance.

2  South Korea is mobilizing $576B in chip and AI investment, with Samsung expecting an 18-fold profit jump this quarter.

President Lee Jae Myung has ordered officials to fast-track permits and infrastructure for a national chip program, with Samsung and SK Hynix each committing $260 billion to new manufacturing sites. The urgency is backed by hard numbers: Samsung is expected to report a record $56B quarterly profit on AI memory demand alone, while SK Hynix, up 273% this year, is launching a $28B Nasdaq listing to fund further expansion.

3  Nvidia launched a program to get tens of thousands of GPUs into AI startups without requiring them to buy chips outright.

The new partnership initiative connects fast-growing AI companies with Nvidia's cloud provider network, trading access to compute for a share of product and cloud revenue. Two Australian firms are first in line: Sharon AI gets 40,000 Nvidia GPUs, and Firmus Technologies is building a data center in Indonesia. For startups stuck in the compute queue, this is a meaningful structural change to how the chips actually flow.

The assumption. The consensus view on AI and employment has been consistent: automation replaces labor, AI accelerates automation, therefore AI costs jobs. That framing has driven most of the public debate for the past three years. A new study suggests it is at least partially wrong, and the mechanism behind the error is worth understanding.

What the data shows. Researchers at Ramp tracked AI spending against hiring data across nearly 22,000 US firms. Companies that adopted AI most aggressively grew their white-collar headcount by 10.2% in the two years following adoption. Entry-level hiring, the category most commonly cited as the first casualty of AI, rose even faster inside those same firms. The gains are real, and they show up in the data consistently enough that they cannot be dismissed as noise.

Three hiring surprises. The labor shift is producing some genuinely unexpected outcomes. AI labs are recruiting philosophy majors to work on questions of model behavior and consciousness, with demand now outpacing supply. Freelancers are billing thousands of dollars to fix AI-generated copy that clients commissioned and then found unusable. And staffing firm Robert Half found that 32% of managers who cut a role for AI later rehired for the same position. Gartner expects half of all AI-attributed cuts to be reversed by 2027.

The honest caveat. The Ramp data clusters heavily around tech and high-spending firms, which were likely already growing before they adopted AI. It is not a clean read on the broader economy, and none of it means AI cannot displace jobs at scale further down the road. What it does mean is that the current picture is considerably messier than the doomsday headlines suggest, and that the businesses spending most aggressively on AI are not, on average, cutting people to pay for it.

America 250, reimagined. Google's July 4th ad asks: what if the Declaration of Independence had been written on a Google Doc? The answer involves a very awkward sharing request from King George III. It is the kind of brand ad that earns its views rather than buying them.

Classic game, new platform. Developer Ammar Reshi ported the 2003 strategy game Command and Conquer to iPhone and iPad using Claude Fable 5. His full breakdown of how it worked is a practical demonstration of what a skilled developer can do with a capable model and a clear brief.

Games from prompts. Meta quietly released an app that lets anyone build AI-generated games from a phone using simple text prompts. Try it here. The barrier to entry for interactive content just dropped considerably.

25 agent loops, ready to run. A Claude Fable 5 loop library packs 25 pre-built agentic workflows covering coding, research, and marketing into a single resource. If you have been meaning to experiment with agent loops but have not started, this is the lowest-friction entry point available right now.

Extract Fable 5 before it goes paid. Claude Fable 5 moves to pay-per-use tomorrow. A viral prompt making the rounds uses a targeted extraction technique to pull Fable 5's reasoning patterns and best practices into a reusable knowledge base, so cheaper models can replicate its approach on future tasks. Worth running today while access is still free.

Design your next 10 years before the year designs them for you

This ChatGPT prompt acts as a future-self coach and transformation strategist. It maps who you want to become across four time horizons, one year, three years, five years, and ten years, across every domain that matters: career, health, finances, relationships, skills, mindset, and purpose. Then it builds a structured roadmap with milestones, habits, and monthly action steps you can actually track. Run it in ChatGPT or Claude and let it ask you questions first before it builds anything.

Let the AI ask its clarifying questions before you fill in any details. The questions it generates are often more useful than the answers, because they surface assumptions about your future that you have never put into words. If you want a sharper output, add one sentence after the prompt describing where you are right now: your current role, your biggest constraint, and the one thing you most want to change in the next 12 months. That context alone tends to tighten the roadmap considerably.

The next wave of millionaires will be people who figured out how to make AI work for them.

The window to get ahead is still open. But not for long.

Here are 200+ proven ways to make money with AI in 2026.

Sign up for Superhuman AI, the free daily newsletter read by 1M+ professionals, and get instant access to all 200+ ways to profit from AI this year.

Claim your free list