Ex-OpenAI researcher cites RLHF limits in AI automation
InstructGPT co-author Diogo Almeida warns that RLHF models suit human assistance, not unsupervised enterprise automation. Here is why current tools fail.
Codex Desktop Adds Free Composer Predictions in Pro Beta
OpenAI added composer predictions to Codex Desktop for ChatGPT Pro users. The beta suggests prompts via Tab without consuming rate limits or credits.
Domyn CEO argues enterprises must own AI models
Domyn CEO Uljan Sharka argues regulated firms must own AI models and data instead of renting commercial APIs from major cloud providers.
Anthropic Launches Claude Dynamic Workflows Beta
Anthropic opened a public beta for Claude dynamic workflows, letting models write code to coordinate multi-agent fleets in the background.
OpenAI Launches Ultrafast Mode for GPT-6.1 Sol
OpenAI launches Ultrafast mode for GPT-6.1 Sol across API, Codex, and ChatGPT Work, charging 6x standard rates for up to eight times faster inference.
Runway and Anthropic Link Claude Motion to Video
Anthropic launched Claude Motion in beta, partnering with Runway to turn deterministic charts and data into animated video for enterprise teams.
Zyphra Trains Reasoning Model on Full AMD MI300X Stack
Zyphra completed pre-training for ZAYA1-8B on an AMD MI300X cluster in IBM Cloud, running without Nvidia hardware or software for up to 72 hours.
SemanTok: Stability tests smaller video world models
Stability AI and KIT introduce SemanTok, using semantic predictability to build compact video world models that outperform much larger baselines.
H Company Speeds VLM Serving with NVIDIA Dynamo
H Company pairs NVIDIA Dynamo with RDMA on AWS, slashing VLM boot times from 27 minutes to 26 seconds for computer-use agents.
Cursor /visualize Updates Charts Within Chat
Cursor updated its /visualize tool to generate follow-up charts in the same chat session, letting developers explore metrics without restarting prompts.