--- title: "OpenAI Unveils GPT-6! Rumored to Have 10 Trillion Parameters, Forced August Release" type: "News" locale: "en" url: "https://longbridge.com/en/news/295339034.md" description: "According to leaks, OpenAI plans to release GPT-6 (codenamed Astra) with 10 trillion parameters in August, a scale far exceeding that of GPT-4. Additionally, an even larger model codenamed Doug is expected to launch by year-end, potentially trained on NVIDIA's next-generation chips. This move highlights OpenAI's massive investment in pre-training resources and its intention to accelerate releases" datetime: "2026-08-10T00:25:40.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/295339034.md) - [en](https://longbridge.com/en/news/295339034.md) - [zh-HK](https://longbridge.com/zh-HK/news/295339034.md) --- # OpenAI Unveils GPT-6! Rumored to Have 10 Trillion Parameters, Forced August Release Just moments ago, shocking news broke on X. AI insider reporter ChrisGPT revealed that despite pressure from government scrutiny, GPT-6 (Astra) will still be released in August. GPT-6 is the model Astra whose release was urgently paused by OpenAI yesterday. Nine months ago, OpenAI began to exert effort, reallocating computing resources. Realizing that Anthropic was leading in pre-training, OpenAI grew anxious! It is estimated that GPT-4 has approximately 1.8 trillion parameters. If Astra truly has 10 trillion parameters, that would be more than five times the size of GPT-4. Regardless, OpenAI has invested an unprecedented scale in pre-training. But wait, there are more heavy hitters coming. The second revelation from ChrisGPT is the year-end "epic beast" codenamed Doug. It is not only the largest pre-trained model OpenAI has ever attempted, but it is also said to make Fable look like a primitive product. Moreover, this model is not GPT-6. However, this model requires pre-training first, followed by extensive cybersecurity testing, so it may be released no later than November. Coincidentally, recently, OpenAI President Greg reposted a post on X— GPT-4 completed its training four years ago on August 8. Industry speculation suggests that while GPT-4 had 1.8 trillion parameters, GPT-6 has reached 10 trillion parameters four years later, achieving a terrifying leap in magnitude! It can be inferred that GPT-6 will emerge with unimaginable logical reasoning and generalization capabilities. ## Year-End Ultimate Beast Doug: Comparing Fable 5 to "Cavemen" The name Doug was not only leaked by ChrisGPT but was also mentioned previously by the well-known institution SemiAnalysis. According to the leaked information, Doug's computational foundation is extremely luxurious. Prominent tech influencer Pankaj Kumar pointed out that Doug is likely being trained using NVIDIA's next-generation Vera Rubin chips, which have not yet been fully deployed! Overall, OpenAI is playing a very large game. In August, it will first use the 10-trillion-parameter GPT-6 to make a splash and attract users; by year-end, it will unleash the beast Doug to secure the absolute top spot. Tibo, who is currently battling Boris, also teased a surprise ## August: The Decisive Battle for ASI Back in July 2026, multiple sources leaked that OpenAI's upcoming next-generation model would adopt a new pre-training foundation (surpassing the previous "Spud" with approximately 4T tokens), with a scale of about 10T and a context window potentially reaching 1.5M tokens, rivaling Anthropic's Mythos/Fable. The situation is now clear: Astra's release was slightly delayed due to a safety pause, but Anthropic has no intention of waiting. Anthropic is preparing to release Fable 5.1 in August, priced similarly to Fable 5, directly competing with Astra. Anthropic is likely employing a "Tian Ji Horse Racing" tactic: Opus 5 was deliberately held back at launch to keep Fable 5.1 as an ambush force, ready to intercept immediately upon GPT-6's debut. Who will become the king of large language models? This question has become unprecedentedly intense in August. ## Breaking Through the "Pre-training Wall": OpenAI's Counterattack Why is Doug so important? Because it points to a reality that the entire industry has been afraid to face: Since the release of GPT-4o in May 2024, OpenAI has not completed full-scale pre-training for the next generation of frontier models for over two years. o1, o3, GPT-5, and GPT-5.5 were essentially post-training, reinforcement learning, and inference computations built on the GPT-4o foundation. The "engine" of GPT has not been replaced for two years. Due to severe pre-training bottlenecks, projects like Orion were downgraded due to underperformance. However! OpenAI has never abandoned the core thread of pre-training. The release of Gemini 3 late last year completely convinced Altman to replace the "engine." Subsequently, the new pre-training model codenamed Garlic emerged. It is reported that Chief Research Officer Mark Chen told the team that the company had resolved key issues in pre-training. But Garlic was more of a verification experiment, while Doug is the true product scaled up significantly. They likely overcame the bottleneck of performance degradation and mastered the stability of clusters with ten thousand or even hundreds of thousands of cards, leading to the birth of Doug. For a long time, SemiAnalysis has been emphasizing that Scaling Laws remain effective. The o1 series proved that models can continuously scale in "inference time." And the return of the king, Doug, declares that in the fundamental dimension of "pre-training," as long as computing power and data keep pace, Scaling Laws can still create miracles! ## Only Two of the Three Giants Remain Now, only two of the "Big Three" remain. On August 5, a corner of Google's AI landscape collapsed. Jeff Dean, the Chief Scientist who served for 27 years, left along with three core researchers to start a new venture. On the same day, Nobel laureate Hassabis stepped down as CEO of DeepMind, retiring to the position of Chairman. Alphabet's stock price fell more than 4% that day. DeepMind is no longer a frontier laboratory. The place that invented the Transformer is now supplying top researchers to competitors. Not long ago, Silicon Valley had the "Big Three" of ASI; now, only OpenAI and Anthropic remain in the race. Currently, Anthropic holds Fable 5. OpenAI holds the sequentially iterated 5.5, 5.6, and the upcoming 6. Now, OpenAI is holding two cards: Astra in August, and Doug at year-end. Over the past two years, OpenAI has proven that the old engine can still squeeze out much more horsepower. When the engine is replaced, how fast can this system, already tuned to its limit, push the car? The answer will be revealed by Doug at year-end. New Zhiyuan Risk Warning and Disclaimer The market carries risks; investment requires caution. This article does not constitute personal investment advice, nor does it consider the specific investment goals, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this content is at your own risk. ### Related Stocks - [OpenAI.NA](https://longbridge.com/en/quote/OpenAI.NA.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) ## Related News & Research - [OpenAI boosts AI chip dreams with AI patents from Altman-backed startup](https://longbridge.com/en/news/295148688.md) - [The Ball in OpenAI's court](https://longbridge.com/en/news/295318599.md) - [ChatGPT brings unlimited text chats to free users](https://longbridge.com/en/news/295146087.md) - [OpenAI has hit back at Apple's lawsuit — with receipts](https://longbridge.com/en/news/294801423.md) - [What smart people are saying about OpenAI's 'Apple is getting this wrong' rebuttal](https://longbridge.com/en/news/294857318.md)