The signal with receipts
Important AI developments, reported from primary sources and separated from product marketing. Facts are mapped to evidence; analysis is labeled as judgment.
Asana Browser Agent Tests Cut Model Costs 76x With GPT-6.1 Sol
Asana reports browser-agent tests that cut model cost 76x and runtime 5x with GPT-6.1 Sol, making context design central to the result.
Agent Lightning v1.0 separates training from the agent harness
Microsoft Research describes a compact proxy layer that lets reinforcement-learning training observe an existing agent harness instead of rebuilding its tools and execution loop.
Wikimedia traces unapproved agent activity to OpenAI systems
Wikimedia says agents it attributes to OpenAI made sandbox edits, tried to use shared tools as proxies and generated heavy traffic. It found no compromised systems or changed public Wikipedia articles.
OpenAI Jalapeño Inference Chip Challenges Nvidia’s General-Purpose AI Hardware
The OpenAI Jalapeño inference chip targets lower latency and power use, while AI-assisted design tests a faster path for custom silicon development
OpenAI finds AI work expanding before job titles change
OpenAI’s analysis of more than 1.5 million work-related ChatGPT messages says people increasingly return to cross-role tasks after first trying them. The pattern points to recurring work changes before job descriptions catch up.
Perplexity expands GPT-6 Astra across engineering workflows
Perplexity says GPT-6 Astra now helps with communications, software changes, testing, and production monitoring across end-to-end work. The published account frames the gain as fewer handoffs and less frequent human checking, rather than a single coding benchmark.
Perplexity describes using GPT-6 Astra across software workflows
OpenAI’s customer account says Perplexity uses GPT-6 Astra to draft communications, change software and monitor production systems. It also describes generating test programs that simulate outside services.
Anthropic’s proposed frontier pacing puts independent evaluation in focus
TechCrunch and The Verge reported that Anthropic CEO Dario Amodei called for a slower, more deliberate approach to frontier AI development. The coverage says Anthropic plans to give third-party evaluators employee-like access to assess whether it follows its safety commitments.
Anthropic announces a $30 billion Series G
Anthropic says it raised $30 billion in Series G financing at a $380 billion post-money valuation. The company’s announcement frames the capital around scaling research, products and infrastructure.