It is code red at OpenAI.

Roughly 2 weeks since the launch of Gemini 3, Sam Altman internally raised a memo to staff due to ChatGPT’s unique daily active users reportedly being down by 6% since mid-November.

There was some internal pushback but the competition is fierce going into the new year so OpenAI decided to get a model out in GPT 5.2 as a Christmas present for us. (unless there’s more)

There are mixed reviews in the initial impressions where it is supposedly too good at complex and autonomous reasoning but only an incremental upgrade for daily use.

Just like GPT 5.1 launched with personalities being the core selling point, GPT 5.2 launched as the most capable model for “professional knowledge work”. So we put it to test.

But before that, quick catchup on the latest news:

1. Remio: Your second brain that doesn't need manual feeding
Runs in background capturing web pages, files, meetings, emails into a local knowledge base. Ask it anything and it pulls from your actual work context instead of giving generic ChatGPT responses.

2. Nume: AI CFO that actually answers anytime
Connect QuickBooks/Xero once, ask about cash flow or expense patterns anytime. Sends Slack alerts when something's off (expense spikes, cash flow shifts).

3. Macaly: Build sites by talking
Describe what you want, sketch on canvas, or just prompt it. Sites, apps, dashboards appear. Handles databases, hosting, SEO, image generation.

The new model comes in two variants: GPT-5.2 and GPT-5.2 Pro. GPT-5.2 Mini nowhere to be seen (yet).

It is available via their UI in both “instant” and “thinking” modes, still corresponding to the API concept of different reasoning effort levels.

The knowledge cut-off date for both variants is now August 31st 2025.

Both of the 5.2 models have a 400,000 token context window and 128,000 max output tokens, so same as 5.1 or 5.

Spreadsheet generation, presentations and other complex data analysis in GPT 5.2 is one of the highlights of the release.

The ability to handle long context with fewer hallucinations is the most important improvement in my opinion.

The Test: A messy e-commerce data and a demanding prompt.

6 months of sales data (Jan-June 2025) with revenue, customers, orders, product categories, and acquisition channels.

Asked for a complete analysis with dashboard, trends, and Q3 projections.

Output:

Generated a complete 6-tab spreadsheet with:

Executive Dashboard:

Trend Analysis:

Q3 Projections:

It captured key insights: Electronics declining from 45% to 33% revenue share while Home & Garden grew from 25% to 37%. AOV increased 14.2% ($239 to $273) but plateaued mid-quarter.

Time: About ~5 minutes. Very slow.

Gemini 3 Pro with the same prompt: Just gave the text analysis and didn’t add the spreadsheet at first.

After following up it created a google sheet but its far less comprehensive.

GPT 5.2 delivers. Far slower but gets the job done.

“Professional knowledge work" angle OpenAI's pushing is legit.

Since launch, developers have been stress-testing 5.2. Here's what's working:

Complex one-shot builds:

Pietro Schirano built a complete 3D graphics engine in a single prompt.

But there’s a big tradeoff:

Speed is a dealbreaker and it’s not looking good for GPT-5.2

GPT 5.2 Thinking mode is the slowest frontier model right now.

Pricing

Mostly evenly priced on web and for API access too.

To conclude I agree with Cristian, this is what we expected GPT 5.0 to be all along. But hey, better late than never.

Try it yourself: Test out longer prompts for creative writing, coding and working with spreadsheets.

Reply with an output that broke your perception about Gemini 3 being the best and got you back onboard with OpenAI products.

Until next time,
Vaibhav 🤝🏻

PS: We’re considering an year in review for AI in 2025. Reply and let us know if there’s anything you’d like us to cover specifically and we’ll get on it.

If you read till here, you might find this interesting

#AD 1

Energy giants like Exxon and Chevron are buying land in America’s lithium hotspot. Now they’ve got a new neighbor. EnergyX has now acquired 35k gross acres in Southwest Arkansas, right next door. With tech that can recover 3X more lithium than traditional methods, they’re aiming for America’s lithium crown. General Motors already invested. Join them and 40k people as an early-stage EnergyX investor today.

Invest in EnergyX Now

_{Energy Exploration Technologies, Inc. (“EnergyX”) has engaged Nice News to publish this communication in connection with EnergyX’s ongoing Regulation A offering. Nice News has been paid in cash and may receive additional compensation. Nice News and/or its affiliates do not currently hold securities of EnergyX.}

_{This compensation and any current or future ownership interest could create a conflict of interest. Please consider this disclosure alongside EnergyX’s offering materials. EnergyX’s Regulation A offering has been qualified by the SEC. Offers and sales may be made only by means of the qualified offering circular. Before investing, carefully review the offering circular, including the risk factors. The offering circular is available at invest.energyx.com/.}

#AD 2

“When I first opened Attio, I instantly got the feeling this was the next generation of CRM.”
— Margaret Shen, Head of GTM at Modal

Attio is the AI-native CRM for modern teams. With automatic enrichment, call intelligence, AI agents, flexible workflows and more, Attio works for any business and only takes minutes to set up.

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉Start for free today