Side-by-side comparisons of language, image and voice models: quality, speed, price, limits and best use.
OpenAI has announced its priorities and guiding principles for third-party safety assessments of frontier AI models. The framework emphasizes rigor, security, independence, and evaluating safeguards b…
Anthropic's Claude Opus 5.5 arrives less than two months after Opus 5, claiming to match Fable 5.1 performance at roughly 40% lower cost. The update brings 20% cheaper tokens, 30% faster output, 20% h…
xAI has released Grok 4.7, a flagship model built on a larger base with longer reinforcement learning training, while keeping Grok 4.6's $2/$6 pricing and speed. It features a 500K-token context windo…
Two new research efforts on reversible generative models have been introduced, each taking a distinct approach to the sa
Cohere, OpenAI, and AI21 Labs have developed a preliminary set of best practices applicable to any organization developi
OpenAI said its DALL·E API is now available in public beta. Developers can begin building apps with the API, and new use
A batch of GPT-focused updates describes model releases, customization options and applications. GPT-3 can be customized
Two related research notes focus on CLIP, a model associated with connecting text and images. One note is titled “Hierar
Cohere, OpenAI, and AI21 Labs have developed a preliminary set of best practices applicable to any organization developi
A neural network called DALL·E creates images from text captions for a wide range of concepts expressible in natural lan
The DALL·E API is now available in public beta, and developers can begin building apps with it starting today. Alongside
The materials include the title “Learning to reason with LLMs” and describe a neural network trained to play Minecraft b
A site note says that fine-tuning GPT-3 can power and scale done-for-you video creation. A closely related note, framed
ChatGPT can now see, hear, and speak. ChatGPT Ads is also expanding to Southeast Asia and Taiwan, giving eligible busine
WHOOP delivers personalized fitness and health coaching with GPT-4, applying a large language model to tailor guidance t
OpenAI has banned accounts linked to Russia's Doppelganger influence operation that used its AI tools to generate anti-Ukraine social media comments, multilingual translations, and website copy. The c…
OpenAI banned an account, likely based in the US, after it used the company's AI tools to generate a fabricated ChatGPT error message falsely claiming 'Russian troll' activity had been detected. The f…
A privacy-protected experiment tested how ChatGPT responds to different names using AI research assistants. Results show no dramatic bias, but subtle shifts in tone, examples, and cultural references …
HealthBench is OpenAI's new benchmark evaluating how AI models perform in realistic healthcare scenarios. Developed with over 250 physicians, it tests accuracy, safety, context awareness, and communic…
Microsoft and OpenAI have announced a new agreement updating their partnership terms, with Azure remaining the primary computing backbone for OpenAI's research and models. The deal blends investment, …
