OpenAI plans to officially release ChatGPT Images 2.5 on Tuesday, September 8, 2026. This version features significant improvements over previous versions in terms of face recognition, pet recognition, and image editing capabilities. According to early tests by Axios, the new engine runs faster, better preserves the similarity of people and pets, and supports more detailed editing operations. Users can input text to add objects or make precise modifications, without relying solely on natural language descriptions. Additionally, security sources indicate that the model has demonstrated the ability to generate realistic conference scenes during testing, such as creating fake GPT-6 conference and Ultraman team content, with no flaws in the text content, natural facial details, and significantly improved noise control.
- 2026-09-07 21:02The `GPT-6 Astra model was released, and OpenAI President Brockman confidently declared, “Welcome to the era of AGI.”
- 2026-09-08 19:21OpenAI released GPT-6 Astra, marking the arrival of the AGI era, with capabilities for autonomous computer operation and scientific research.
- 2026-09-08 22:40① Nvidia CEO Jensen Huang posted that OpenAI’s GPT-6 Astra, released last week, was trained using approximately 100,000 NV Link 72 clusters, and he believes that General Artificial Intelligence (AGI) has officially arrived; ② GPT-6 Astra can directly operate computers and software to perform complex tasks such as programming, reaching the most advanced level in multiple fields. OpenAI has announced the beginning of the AGI era.
Event frameEVENT FRAME
国内 · 主流媒体across 3 days
OpenAI plans to officially release ChatGPT Images 2.5 on September 8, 2026.
Coverage · reports per dayLANGUAGE SPLIT
Integrated timelineUNIFIED TIMELINE
-
2026-09-03
Leakage and testing progress of ChatGPT Images 2.5
According to Qwant, the new GPT Image 2.5 can simulate the GPT-6 launch event, indicating significant upgrades. A new pop-up message has appeared in ChatGPT, confirming the version number as 2.5 instead of the speculated 2.1.
-
2026-09-04
Leakage and testing progress of ChatGPT Images 2.5
According to security insider reports, the new GPT Image 2.5 can simulate the GPT-6 launch event, showing significant upgrades. A new pop-up message has appeared in ChatGPT, confirming the version number as 2.5 instead of the speculated 2.1.
-
2026-09-08
Official release of ChatGPT Images 2.5
OpenAI released ChatGPT Images 2.5 on Tuesday, with significant improvements in face recognition, pets, and editing capabilities. The new engine runs faster, retains similarity, and supports detailed text editing. Tests show that it can generate realistic launch event scenes.
SignalsSIGNALS
All reports (3)SOURCES
The new version of GPT Image 2.5 can already fake the GPT-6 launch event.
# The new GPT Image 2.5 can now fake the GPT-6 launch event The host of the GPT-6 launch event is actually Cyber Father Tibo, but this is just a fake version of the new image generation model GPT Image 2.5. Although it hasn’t been officially released yet, new pop-up alerts have appeared in ChatGPT. Unlike the previously speculated version named GPT Image 2.1, the new version’s version number is clearly 2.5, indicating a significant upgrade. According to test results, GPT Image 2.5 can directly fake content such as the official GPT-6 launch event and the full list of Ultramen characters. It can generate screenshots of Ultramen posts, with no flaws in the text content aside from the avatar. During testing, the model generated images very quickly, with realistic photo quality. Even when multiple images of the same character were generated, the facial details remained natural, and the embedded text also did not deform easily. Compared to GPT Image 2, the new version has significantly improved noise control. Reference link: [1]https://x.com/chetas…
ChatGPT Images gets better at faces, pets and edits
OpenAI 将于周二发布 ChatGPT Images 2.5,该版本在人脸、宠物及编辑功能上较前代有显著提升。根据 Axios 的早期测试,新引擎不仅运行更快,还能更好地保留人物和宠物的 likeness(相似性),并支持更详细的编辑操作。Axios 团队通过多项任务验证了性能,包括为 Raven 生成图像、将猫照转换为不同风格以及设计纹身图案。此外,新增工具允许用户输入文字添加对象或进行精确修改,而无需仅依赖自然语言描述。OpenAI 产品主管 Adele Li 表示,这些功能旨在让用户更主动地参与创作过程。