OpenAI has been quite active this year, launching a variety of thrilling AI products, including Operator and its inaugural AI agent, as well as the new 4o image generator model integrated into ChatGPT and the o3 reasoning chatbot. The newest entry is the enhanced ChatGPT Agent, which was released last week, and OpenAI is already witnessing a spike in usage. The ChatGPT Agent experienced considerable interest from Pro subscribers, prompting OpenAI to delay the rollout for the Plus and Team tiers by a few days.
However, ChatGPT users continue to await OpenAI’s most significant upgrades of the year. To begin with, this includes the GPT-5 model that is anticipated to debut this summer. Additionally, there is the open-source ChatGPT version, which OpenAI has postponed twice in recent months.
GPT-5 is undeniably the most thrilling OpenAI product of the year as it is expected to enhance the overall chatbot experience. GPT-5 is likely to succeed GPT-4o, the model that the majority of users employ for their interactions with ChatGPT. Notably, there has been an increase in teasers for GPT-5 lately, including a leak indicating that GPT-5 is currently undergoing testing.
OpenAI CEO Sam Altman participated in the ChatGPT Agent livestream last week. Subsequently, on Saturday morning, he took to X to hint at an unreleased ChatGPT product. “Woke up early on a Saturday to have a couple of hours to try using our new model for a little coding project,” Altman remarked. “Done in 5 minutes. It is very, very good. Not sure how I feel about it…”
This doesn’t definitively confirm that Altman was utilizing GPT-5 for his coding project. Given that vibe coding is one of the AI features that has garnered considerable interest over the last few years, we anticipate GPT-5 to showcase more sophisticated coding capabilities. It would make sense for Altman to test GPT-5 prior to its public launch. After all, the CEO mentioned in the spring that GPT-5 would be released in the summer.
Hours before Altman tweeted, prominent X user Tibor Blaho shared a screenshot featuring a GPT-5 teaser from Alexander Wei in an X thread where the OpenAI engineer described how a new OpenAI reasoning model achieved gold medal-level performance at the International Math Olympiad. That teaser is still accessible on X, stating: “btw, we are releasing GPT-5 soon, and we’re excited for you to try it. But just to be clear: the IMO gold LLM is an experimental research model. We don’t plan to release anything with this level of math capability for several months.”
Blaho frequently uncovers information regarding unreleased AI models, sharing it on X before the products become commercially available. A few hours after sharing Wei’s teaser, but before Altman’s cryptic tweet, Blaho discovered evidence of GPT-5 being tested in the open. An OpenAI model referred to as “openai/gpt-5-reasoning-alpha-2025-07-13” is already undergoing testing. The designation for this GPT-5 model suggests it is a reasoning AI chatbot, akin to ChatGPT o3 and o4-mini that are presently accessible to users.
Given Altman’s prior comments regarding OpenAI’s plans for ChatGPT, we expect GPT-5 to be the first version where users will not have to choose the type of experience desired. The AI should possess the intelligence to automatically transition to reasoning when required to complete a task. Nevertheless, OpenAI might employ different models under the GPT-5 banner to achieve this performance, potentially having distinct names internally.
All of this remains speculative, as OpenAI has not disclosed a definitive timeline for the GPT-5 release. Additionally, delays can always occur, as seen with the open-source model. More than an hour after Altman’s teaser, OpenAI shared a similar teaser on its X account. “To be clear: We’re releasing GPT-5 soon, but the model we used at IMO is a separate experimental model,” OpenAI stated. “It utilizes new research methodologies that will manifest in future models—but we don’t intend to release a model with this level of capability for many months.”