All briefings and quick hits published that day.
OpenAI says its first inference chip delivers lower latency and more work per watt across three large open-weight models.
It is the clearest evidence yet that a frontier-model company may materially weaken Nvidia’s hold on large-scale inference.
Portable Computer keeps agent planning, file analysis and tool execution on a DGX Spark, escalating to cloud models only with permission.
EA and three global music groups joined Stability AI’s Series B, tying its next phase more closely to licensed creative production.
Gradio’s new Workflow interface turns typed AI pipelines into inspectable canvases, callable REST endpoints and deployable applications.
OpenAI banned accounts tied to a covert Russian campaign that used ChatGPT to promote a fabricated institute and stolen scholarship.