Two Years, Every Month a Headline: AI From Sora to GPT-5 and Nano Banana
Faisal Adnan · 05 Nov 2026
February 2024: Sora
OpenAI showed Sora, a model that turns text into video up to a minute long. The famous demo prompt described a stylish woman walking down a Tokyo street. Axios covered the announcement.
October 2024: an AI Nobel
Demis Hassabis and John Jumper of Google DeepMind were co-awarded the 2024 Nobel Prize in Chemistry for AlphaFold, which predicts the 3D structure of proteins from their amino acid sequences. The system has predicted around 200 million structures, which is work that used to take a researcher a whole PhD for one protein.
January 2025: the DeepSeek shock
A low-cost Chinese model from DeepSeek matched OpenAI's o1 on many tests. The market reacted hard: according to Reuters, as republished by the Kathmandu Post, Nvidia posted a record market-value loss of around $593 billion in a single day. Investors worried that a cheap model could threaten the leaders.
March 2025: the Ghibli trend
Everyone turned their selfies into Studio Ghibli style art. Sam Altman posted that "our GPUs are melting" and OpenAI introduced temporary rate limits. Fortune also noted the copyright questions the trend raised.
May 2025: Veo 3
Google's Veo 3 generated video together with sound: dialogue, effects and music. TechCrunch described it as a model that can create audio to go along with its clips. For AI video, the silent era was over.
August 2025: GPT-5
OpenAI released GPT-5, which was pitched as a PhD-level expert in your pocket, and PBS NewsHour noted how high the expectations were. It was available to free users, too.
August and September 2025: Nano Banana
Google's updated image editing model in the Gemini app focused on keeping a person's or pet's likeness consistent across edits. More than 5 billion images were made with it in under a month, and Gemini reached number one on the App Store.
September 2025: the Sora 2 app
OpenAI launched an app that works like a TikTok of AI-generated video. TechCrunch reported it reached number one on Apple's US App Store while still invite-only, with around a million downloads in under five days.
What I want you to take away
- The pace is the story. Skills and tools you learn today may be replaced within a year, so learn the principles and keep practising.
- Competition is global. A low-cost model from one lab can shake an entire market in one day.
- Every viral moment has a cost: compute, copyright questions, or deepfakes. Notice who pays.
Sources
- Axios: OpenAI's Sora text-to-video
- Google: Hassabis and Jumper awarded the Nobel Prize in Chemistry
- Reuters via Kathmandu Post: DeepSeek sparks AI stock selloff
- Fortune: Altman says Ghibli-style images are melting OpenAI's GPUs
- TechCrunch: Veo 3 can generate videos and soundtracks
- PBS NewsHour: OpenAI releases GPT-5
- Google: Nano Banana image editing in Gemini
- TechCrunch: Sora soars to No. 1 on Apple's US App Store
- Google DeepMind: Veo and OpenAI: Sora (demo samples)
Part of the talk "The AI Era" by Faisal Adnan, Riphah International University, 6 November 2026.