OpenAIIntroducing ChatGPT Images 2.0
A concise technical overview of ChatGPT Images 2.0, highlighting the new era of image generation, available modes (Image mode and Classic mode), and various format options.
Google DeepMindPartnering with industry leaders to accelerate AI transformation
Explores how Google DeepMind partners with Accenture, Bain & Company, BCG, Deloitte, and McKinsey to accelerate AI-driven enterprise transformation through frontier AI models, industry-specific capabilities, and responsible, leadership-guided deployment at scale.
OpenAIOpenAI helps Hyatt advance AI among colleagues
Hyatt scales AI-powered guest experiences and cross-department productivity by deploying ChatGPT Enterprise with OpenAI, enabling live onboarding and data-driven collaboration across finance, marketing, operations, and customer-facing teams.
OpenAIOpenAI helps Hyatt advance AI among colleagues
Hyatt scales enterprise AI with OpenAI via ChatGPT Enterprise, deploying frontier capabilities (GPT-5.4, Codex) across finance, marketing, operations, and guest-facing teams to boost productivity, streamline workflows, and deliver more personalized guest experiences through live onboarding and broad employee AI access.
OpenAIOpenAI helps Hyatt advance AI among colleagues
Hyatt scales enterprise AI by deploying ChatGPT Enterprise across its global workforce, leveraging OpenAI tools such as GPT-5.4 and Codex to boost productivity, streamline operations, and deliver more personalized guest experiences through AI-powered workflows.
PinterestSmarter URL Normalization at Scale: How MIQPS Powers Content Deduplication at Pinterest
A data-driven MIQPS system learns which URL query parameters matter for content identity on a per-domain basis, enabling scalable, multi-layered URL normalization and content deduplication at Pinterest.
PinterestSmarter URL Normalization at Scale: How MIQPS Powers Content Deduplication at Pinterest
Dynamic, per-domain MIQPS-based URL parameter analysis enables scalable content deduplication and robust normalization at Pinterest.
Berkeley AIGradient-based Planning for World Models at Longer Horizons
GRASP delivers a gradient-based, lifted-state planner for long-horizon world-model control, introducing stochastic exploration and gradient shaping to stabilize planning in high-dimensional differentiable simulators.
Berkeley AIGradient-based Planning for World Models at Longer Horizons
GRASP is a gradient-based, lifted-state planner for learned world models that enables robust long-horizon planning by parallelizing dynamics with a collocation objective, injecting noise into state updates for exploration, and reshaping gradients to avoid brittle state-input sensitivity while periodically syncing with true rollouts.
Ink and SwitchImplementing a Local-First Task Framework
Concrete walkthrough of implementing a Local-First Task Framework, detailing task queues, workers, routers, and takeover coordination powered by Automerge documents.
OpenAIAccelerating the cyber defense ecosystem that protects us all
Accelerating the cyber defense ecosystem by scaling trusted, safeguarded frontier AI capabilities through the Trusted Access for Cyber program, uniting open-source security teams, vulnerability researchers, and enterprises to strengthen software supply chains and critical infrastructure.
OpenAIIntroducing GPT-Rosalind for life sciences research
Introducing GPT-Rosalind, a purpose-built AI that accelerates life sciences research—from target discovery and drug development to evidence synthesis and experimental planning—by enhancing multi-step workflows, tool use, and data integration.