10 results on this page
OpenAI has slowed parts of its AI development to strengthen security and safeguards. The company paused reinforcement-learning training for two weeks on its latest deployment-focused models and delayed its largest planned frontier RL run.
OpenAI has introduced a monitoring system for its riskiest AI work and says it adds about 20% to the compute cost of covered workloads. The company also paused some frontier training for two weeks as part of the changes.
Ornith-1.5 launches open models in 397B, 35B, and 9B parameter sizes. The models feature self-improving task and scaffold generation and report strong performance on coding and reasoning benchmarks.
Google DeepMind announced that the Gemma model family has surpassed one billion global downloads. Developers have released more than 100,000 Gemma variants over the past two years, contributing to an expanding ecosystem called the “Gemmaverse.”
OpenAI released a ChatGPT for macOS plugin that integrates with Apple’s Messages app. Users can search past conversations, draft messages, and send them from Codex or ChatGPT Work chats, with approval required before sending.
Vercel CLI now supports the full Vercel Toolbar comment triage workflow from the terminal. Developers can list, inspect, reply to, resolve, reopen, edit, and delete comments, with JSON output available for scripts and coding agents.
Amid lawsuits, executive departures, and preparations for an IPO, Greg Brockman has emerged as an increasingly powerful figure at OpenAI. The article analyzes how the co-founder and current president has consolidated influence over the company’s engineering and strategic direction.
Entropic Scree is an open-source, model-agnostic diagnostic that uses Normalized Mutual Information to estimate intrinsic rank in complex tabular datasets. It aims to reduce the overestimation and structural failures associated with PCA, Kernel PCA, and Euclidean nearest-neighbor estimators.
Moonshot and ZAI are reportedly approaching the performance of models from OpenAI and Anthropic. The models can reportedly build passable websites at a cost 75% lower than Claude, highlighting potential advantages in cost-sensitive coding workflows.
A position paper analyzing 500 Hugging Face model cards argues that model cards alone do not provide enough information for governing open-weight foundation models. It proposes combining model cards with acceptable use policies and licenses to address safety, provenance, behavior, and enforcement gaps.