Two Years After Sora's Stumble, AI Video Builds a Studio in Hollywood
Promise blends live-action with real-time backgrounds using Seedance 2.5, while Hicksfield's revenue jumps 35x as the industry resets
The company behind ChatGPT and the GPT family of models. It sparked the current AI boom with ChatGPT's launch in 2022. As of August 2026, its flagship model lineup is GPT-5.6, with an improved Sol for paying users and Luna assigned to free users. Around the same time, it released a new model called Astra, which came with warnings about potentially catastrophic cyber capabilities. The company also published an agent plugin standard together with AWS, Cursor, GitHub Copilot, and others. It unveiled a GPT-Live design for seamless voice conversations, and announced plans for a smart speaker slated for release in 2027.
Current rank (1M)
1
Rank over the last 30 days
2026.08.18 – 2026.09.16 · High 1 · Low 2 · Now 1
Promise blends live-action with real-time backgrounds using Seedance 2.5, while Hicksfield's revenue jumps 35x as the industry resets
Former OpenAI staffer Brundage tells Bloomberg third-party audits are needed
Startup valued at $1.3 billion 82 days ago sees valuation more than quintuple under Stripe
LabLLM supports GPT-style model design, training, and chat locally on Apple Silicon via MLX
A Silicon Valley investor and Anthropic's CEO clashed on X over AI regulation
Preparedness team eliminated; bio and cyber risk evaluation moved to existing departments amid internal safety concerns
New custom benchmarking platform compares not just quality but cost and time per task
Under investor pressure, guarantee cut from $25B to under $12B; Anthropic counters bubble talk with 2.4x quarterly revenue jump
Moonshot AI's new benchmark ranks GPT-5.6 Sol first at 59.7%, revealing that many "reasoning errors" are actually perception failures
Small features—automatic quiz generation from conversations, reservation search, and a Google Drive file library—were added all at once
Anthropic has begun embedding invisible markers in text written by Claude. Rather than secretly inserting characters, the technology turns the "dice roll" used to pick words into a secret key — a technique in the same family as Google DeepMind's SynthID. Anthropic switched it on worldwide, citing compliance with the EU AI Act as the reason. On the same day, Google did the opposite, opening up an option to turn off its visible watermarks.
Epoch AI survey finds only science and tech roles get employer-paid subscriptions