GPT-5

On December 2nd, DeepSeek V3.2 was released in the official edition, enhancing Agent's ability to integrate thinking and reasoning. Two official versions of the model were published today: DeepSeek-V3.2 and DeepSeek-V3.2-Special. Official web pages, App and API have been updated to the official DeepSeek-V3.2. The Speciale version is currently available only in the form of a temporary API service for community assessment and research. The new model technical report has been published simultaneously

On November 15, according to the news, yesterday, November 14th, the technology media published an article in which it reported that Mask planned to postpone the launch of the Grok 5 model to 2026, which would have 6 trillion parameters, twice the size of the current Grok 3 and Grok 4. This Friday, at Baron Capital's annual investment conference, Elon Musk did not hide confidence in the Grok 5 model..

According to news from November 9, in September this year, OpenAI launched GPT-5-Codex, a GPT-5 model that optimizes “auto-code” missions on the Codex platform, based on the GPT-5 architecture, which has significantly improved its reasoning and programming capabilities. GPT-5-Codex is a real software engineering landscape capable of working from creating new projects, adding functionality and testing to large-scale code re-engineering. According to Neowin, OpenAI released GPT-5-Code..

Recently, a user named Lex (@xw33btv) wrote on platform X, claiming that OpenAI had entered a new model called GPT-5-Chat-Safety, which did not appear in FAQ, API documents or service terms. It is described that when the user enters an emotional background in GPT-4o, the original response is ignored and replaced with a version produced by "GPT-5-Chat-Safety". Whether it's "I had a bad day" or "I love you."

In the news of September 25th, the technological media blepingcomputer today (25 September) published an article in which it reported that OpenAI was testing the AI smart body called “GPT-Alpha” internally, based on the GPT-5 model, for advanced reasoning and tool use. Netizen @ClayMalott released a tweet today on platform X, sharing a screenshot of a feature called "Agent with Transition" in Alpha Models..

On September 15th, OpenAI claimed that GPT-5 had doctoral skills, but Google DeepMind did not seem to accept such claims. Google DeepMind CEO Demis Hassabis claims that the AIS has developed a system with a doctoral intelligence, which is “nonsense”. As Hathabis said at the All In Summit, "We often hear competitors say that these modern systems we have today..

OpenAI CEO Sam Altman spoke with the editors of tech site TheVerge over dinner about a number of topics, including criticism of GPT-5, the AI bubble, brain-computer interfaces, and the acquisition of Google Chrome. GPT-5 Criticism First, Altman responded to some of the criticisms that followed the release of GPT-5 last week. Just about an hour before Altman's dinner with TheVerge editors, OpenAI pushed out an update that brought back...

Aug. 17 - Even though ChatGPT has introduced a new GPT-5 model, it still has the potential for error, a senior OpenAI executive re-emphasized this week, as Nick Turley, head of OpenAI ChatGPT, noted in an interview with The Verge's Decoder podcast: "In terms of reliability, there's a big discontinuity between reliable and completely reliable." He further explained, "Unless we can demonstrate that in all areas, not just...

ChatGPT recently upgraded to GPT-5, not to consider those professional and technical performance analysis, only on the daily use, has begun to frequently use in-depth analysis for reply, the direct feeling is that the reply is more specific, professional, less illusion, higher comprehension, the previous need for multiple rounds of questioning, usually once you can get a satisfactory answer. This article would like to talk about the simple application of AI in the field of stock investment, the past AI in the stock analysis, unless you have a very professional and exhaustive question, you can get a better result, or else the quick reply is to directly generate a number of similar textbook answers to the words of the piece of soup,...

On August 13, OpenAI CEO Sam Altman announced a major update to ChatGPT. The update introduces user-selectable modes for GPT-5, including Auto, Fast, and Thinking. Thinking.) 1AI notes that Altman said on social media that users can now choose between the different modes of GPT-5. Most users are likely to choose "Auto" mode, but for those who need extra control...

Aug. 12 (Bloomberg) -- OpenAI last week officially launched its next-generation artificial intelligence model, GPT-5, which has filed trademark applications in China under the international classifications of scientific instruments and website services, all of which have been rejected so far. Specifically, OpenAI has filed trademark applications in China through two affiliates, of which: OPENAI has applied to register two "OPENAI GPT-5" trademarks, with international classifications covering website services and scientific instruments, both of which are currently awaiting refusal review; and OPENAI Operations LLC has applied to register two "GPT-5" trademarks, with international classifications covering website services and scientific instruments. OPENAI applied for the registration of two "GPT...

The performance of GPT-5 since its launch has sparked a huge controversy, being accused of lowering the IQ decline, and GPT-4o has also strongly called for its return. A few days ago, in the well-known Norwegian Mensa IQ test, GPT-5 set the worst record in the model: GPT-5 Thinking scored 85 in the test, and 57 in the offline test, directly at the bottom. GPT-5 scored 118 and 70 in the offline test. But the test was not an official Mensa test for AIs, it was a test of 35 graphical reasoning questions that Mensa Norway has made freely available to the public (t...

Aug. 11 (Bloomberg) -- Seventy-two hours after a public revolt by its most loyal users, OpenAI has been forced to drastically adjust its strategy. In a post to X on Sunday, OpenAI CEO Ortmann announced that the company would be restoring older versions of AI models, including GPT-4o, and dramatically increasing access for paid subscribers in a gesture of goodwill to an angry user base. Oltman said early this morning that the percentage of daily use of the inference model has risen significantly: from less than 1% to 7% for free users and from 7% to 24% for Plus users, so raising...

August 11 - Microsoft's Copilot app for Windows 11 (and Windows 10) now supports Smart Mode powered by GPT-5. The feature has been rolling out since August 7 and is now widely available in the U.S. and elsewhere. Windows Latest testing has found that the write box in the Copilot app automatically recognizes new models through the "network routing" feature. While the Copilot app itself is a native app that primarily uses WinUI (...

August 9 news, according to foreign media 9to5mac reported on the 8th, Apple said, Apple Intelligence in the ChatGPT integration will be upgraded to the latest GPT-5 model in iOS 26, iPadOS 26 and macOS Tahoe 26, is expected to roll out as soon as next month. Currently, the Apple Intelligence integration in iOS 18, iPadOS 18, macOS Sequoia and visionOS 2...

August 8, 2012 - OpenAI officially announced the GPT-5 model at a conference held at 1:00 a.m. Beijing time today. OpenAI CEO Sam Altman was the first to take the stage and announced the GPT-5 model in a nutshell, saying that it was an important step towards general artificial intelligence (AGI). According to him, talking to GPT-5 is like talking to a human expert. The OpenAI team took the stage to introduce the GPT-5 model, stating that GPT-5 is already in the pack...

August 5, 2012 - OpenAI engineer Steven Heidel posted early this morning that this week "is going to be an exciting one!" This is going to be an exciting week! According to The Verge, OpenAI is expected to launch GPT-5 this month, which will continue the strategy of its predecessor, with mini and nano versions available via an API. 1AI notes that OpenAI CEO Sam Altman has also been releasing frequent teaser messages recently,...

July 31 news, technology media NeoWin yesterday (July 30) published a blog post, reported that Microsoft plans to introduce "intelligent" mode in Copilot, without the need for users to manually switch, will be based on the task, automatically adjust the speed and depth of thinking. Note: Copilot currently offers the following three modes: Quick Response: response time of about 2~3 seconds, can quickly answer the user's daily questions; more in-depth thinking: response time of 30 seconds, mainly suitable for more complex tasks; contemplative study: response time of 10 minutes ...

A few days ago, according to The Information reporter Stephanie Palazzolo, OpenAI's GPT-5 model has been improved in a number of areas, but according to the source, its performance is particularly strong in programming. Stephanie noted that GPT-5 not only performs better on academic tasks, but also on real-world programming tasks such as working with large and complex code bases. It also said that if GPT-5's performance on these more practical everyday programming tasks improves significantly, it would be a huge advantage to competitor Anth...

July 20, 2012 - GPT-5 is expected to make its official debut in the coming days or weeks. There are indications that OpenAI is testing a new model codenamed gpt-5-reasoning-alpha-2025-07-13. The model was finalized on July 13 and is currently in the final testing phase. According to a code snippet shared tonight by X networker Tibor Blaho, "Models: openai/gpt-5-reasoning-alpha-2025...

On July 17th, OpenAI announced via its X account that it will be broadcasting a live stream on July 17th at 10:00 a.m. Pacific Time (note: 1:00 a.m. BST on July 18th). In the official teaser image released, a mouse pointer shows a pentagonal trajectory, causing netizens to look forward to the latest model GPT-5. According to 1AI's previous report, OpenAI CEO Olbermann personally disclosed the timeline for the release of GPT-5 in a podcast, further reiterating that the "strongest model" will be released this summer...

Early this morning, OpenAI released a 40-minute in-depth interview with its co-founder and CEO Sam Altman. The interview is very technical, Altman talked about the core product GPT-5, which is of great interest to everyone, and is likely to be released this summer, but will also be extended due to naming, security testing, and feature iteration. According to Business Insider, GPT-5 will be far superior to GPT-4, with early testers calling it "a significant improvement in a number of areas". OpenAI's main revenue currently comes from enterprise customers buying...

June 4 (Bloomberg) -- At the AI Summit in Mexico, two OpenAI representatives revealed that the company is developing its next-generation base model, GPT-5, and plans to use it to compete more closely with competitors such as Gemini 2.5 Pro and Claude 4. According to the blog post cited by 1AI, the two OpenAI representatives made it clear that GPT-5 is on the horizon and will far outperform existing models such as GPT-4. They admitted that the cost of development is not yet known, but hinted that the price could be quite high. ...

On May 17th, OpenAI released the software engineering intelligence agent "Codex", and then held an online "Q&A" on Reddit. When asked what GPT-5 can do for users in the future, Jerry Tworek, VP of Research at OpenAI, replied, "GPT-5 is the next generation of the base model, and the team hopes that users will be able to reduce the number of model switches. At the same time, Jerry also mentioned the AI Agent "Oper...