OpenAI fires back at Google with GPT-5.2 after ‘code red’ memo

OpenAI launched its newest frontier mannequin, GPT-5.2, on Thursday amid growing competitors from Google, pitching it as its most superior mannequin but and one designed for builders and on a regular basis skilled use. 

OpenAI’s GPT-5.2 is coming to ChatGPT paid customers and builders by way of the API in three flavors: On the spot, a speed-optimized mannequin for routine queries like information-seeking, writing, and translation; Pondering, which excels at complicated structured work like coding, analyzing lengthy paperwork, math, and planning; and Professional, the top-end mannequin aimed toward delivering most accuracy and reliability for tough issues. 

“We designed 5.2 to unlock much more financial worth for individuals,” Fidji Simo, OpenAI’s chief product officer, stated Thursday throughout a briefing with journalists. “It’s higher at creating spreadsheets, constructing displays, writing code, perceiving photos, understanding lengthy context, utilizing instruments after which linking complicated, multi-step tasks.”

GPT-5.2 lands in the course of an arms race with Google’s Gemini 3, which is topping LMArena’s leaderboard throughout most benchmarks (aside from coding — which Anthropic’s Claude Opus-4.5 nonetheless has on lock).

Early this month, The Data reported that CEO Sam Altman launched an inside “code pink” memo to employees amid ChatGPT site visitors decline and considerations that it’s dropping client market share to Google. The code pink known as for a shift in priorities, together with stalling on commitments like introducing adverts and as a substitute specializing in creating a greater ChatGPT expertise. 

GPT-5.2 is OpenAI’s push to reclaim management, at the same time as some staff reportedly requested for the mannequin launch to be pushed again so the corporate might have extra time to enhance it. And regardless of indications that OpenAI would focus its consideration on client use circumstances by including extra personalization and customization to ChatGPT, the launch of GPT-5.2 seems to beef up its enterprise alternatives. 

The corporate is particularly concentrating on builders and the tooling ecosystem, aiming to grow to be the default basis for constructing AI-powered purposes. Earlier this week, OpenAI launched new information exhibiting enterprise utilization of its AI instruments has surged dramatically over the previous 12 months. 

This comes as Gemini 3 has grow to be tightly built-in into Google’s product and cloud ecosystem for multimodal and agentic workflows. Google this week launched managed MCP servers that make its Google and Cloud companies like Maps and BigQuery simpler for brokers to plug into. (MCPs are the connectors between AI programs and information and instruments.)

OpenAI says GPT-5.2 units new benchmark scores in coding, math, science, imaginative and prescient, long-context reasoning, and gear use, which the corporate claims might result in “extra dependable agentic workflows, production-grade code, and complicated programs that function throughout giant contexts and real-world information.”

These capabilities put it in direct competitors with Gemini 3’s Deep Assume mode, which has been touted as a significant reasoning development concentrating on math, logic, and science. On OpenAI’s personal benchmark chart, GPT-5.2 Pondering edges out Gemini 3 and Anthropic’s Claude Opus 4.5 in almost each listed reasoning check, from real-world software program engineering duties (SWE-Bench Professional) and doctoral-level science data (GPQA Diamond) to summary reasoning and sample discovery (ARC-AGI suites). 

Analysis lead Aidan Clark stated that stronger math scores aren’t nearly fixing equations. Mathematical reasoning, he defined, is a proxy for whether or not a mannequin can comply with multi-step logic, hold numbers constant over time, and keep away from refined errors that would compound over time. 

“These are all properties that actually matter throughout a variety of various workloads,” Clark stated. “Issues like monetary modeling, forecasting, doing an evaluation of knowledge.”

Throughout the briefing, OpenAI product lead Max Schwarzer stated GPT-5.2 “makes substantial enhancements to code era and debugging” and might stroll via complicated math and logic step-by-step. Coding startups like Windsurf and CharlieCode, he added, report “state-of-the-art agent coding efficiency” and measurable good points on complicated multi-step workflows.

Past coding, Schwarzer stated that GPT-5.2 Pondering responses comprise 38% fewer errors than its predecessor, making the mannequin extra reliable for day-to-day decision-making, analysis, and writing. 

GPT-5.2 seems to be much less a reinvention and extra of a consolidation of OpenAI’s final two upgrades. GPT-5, which dropped in August, was a reset that laid the groundwork for a unified system with a router to toggle the mannequin between a quick default mannequin and a deeper “Pondering” mode. November’s GPT-5.1 centered on making that system hotter, extra conversational, and higher suited to agentic and coding duties. The most recent mannequin, GPT-5.2, appears to show up the dial on all of these developments, making it a extra dependable basis for manufacturing use. 

For OpenAI, the stakes have by no means been greater. The corporate has made commitments to the tune of $1.4 trillion for AI infrastructure buildouts over the subsequent few years to help its progress — commitments it made when it nonetheless had the first-mover benefit amongst AI corporations. However now that Google, which lagged behind firstly, is pushing forward, that guess may be what’s driving Altman’s “code pink.” 

OpenAI’s renewed deal with reasoning fashions can be a dangerous flex. The programs behind its Pondering and Deep Analysis modes are costlier to run than normal chatbots as a result of they chew via extra compute. By doubling down on that sort of mannequin with GPT-5.2, OpenAI could also be organising a vicious cycle: spend extra on compute to win the leaderboard, then spend much more to maintain these high-cost fashions working at scale.

OpenAI is already reportedly spending extra on compute than it had beforehand let on. As Trendster reported lately, most of OpenAI’s inference spend — the cash it spends on compute to run a skilled AI mannequin — is being paid in money reasonably than via cloud credit, suggesting the corporate’s compute prices have grown past what partnerships and credit can subsidize.

Throughout the name, Simo urged that as OpenAI scales, it is ready to supply extra services and products to generate extra income to pay for added compute.

“However I believe it’s essential to put that within the grand arc of effectivity,” Simo stated. “You might be getting, right now, much more intelligence for a similar quantity of compute and the identical quantity of {dollars} as you have been a 12 months in the past.”

For all its deal with reasoning, one factor that’s absent from right now’s launch is a brand new picture generator. Altman reportedly stated in his code pink memo that picture era could be a key precedence shifting ahead, significantly after Google’s Nano Banana (the nickname for Google’s Gemini 2.5 Flash Picture mannequin) had a viral second following its August launch.

Final month, Google launched Nano Banana Professional (aka Gemini 3 Professional Picture), an upgraded model with even higher textual content rendering, world data, and an eerie, real-life, unedited vibe to its pictures. It additionally integrates higher throughout Google’s merchandise, as demonstrated over the previous week because it pops up in instruments and workflows like Google Labs Mixboard for automated presentation era.

OpenAI reportedly plans to launch one other new mannequin in January with higher photos, improved pace, and higher persona, although the corporate didn’t verify these plans Thursday.

OpenAI additionally stated Thursday it’s rolling out new security measures round psychological well being use and age verification for teenagers, however didn’t spend a lot of the launch pitching these adjustments.

This text has been up to date with extra details about OpenAI’s compute effectivity standing.

Acquired a delicate tip or confidential paperwork? We’re reporting on the interior workings of the AI business — from the businesses shaping its future to the individuals impacted by their selections. Attain out to Rebecca Bellan at rebecca.bellan@techcrunch.com or Russell Brandom at russell.brandom@techcrunch.com. For safe communication, you possibly can contact them by way of Sign at @rebeccabellan.491 and russellbrandom.49.