Claude Code 2.1.267 adds maxEffortLevel cap across providers
The release ships 53 CLI changes, led by a maxEffortLevel setting (top-level or per model under modelSettings) that caps effort on Bedrock, Vertex and Foundry while letting users pick lower. A --system-prompt-snapshot off flag re-renders the system prompt each request for prompt iteration.
4 sources
More stories today
Skild AI's S1 robot model learns tasks from one video
Skild AI's S1 robot foundation model learns unseen, long-horizon tasks from a single video demonstration via in-context learning, with no weight updates or task-specific post-training. Skild has raised nearly $1.7B since 2023 and hit a $100M annual revenue run rate 10 months after first commercial deployment.
Nvidia AI Blog·57 minutes ago
Tako Search free on Vercel AI Gateway through September 30
Tako Search is a model-agnostic search tool usable with any model on AI Gateway, with no separate Tako account or API key required. It combines a curated knowledge graph with live web results, returning source-grounded data cards, citations, and visualizations.
Vercel Blog·1 hour ago
Google previews Gemini API docs inside AI Studio
Google AI Studio·1 hour ago
Bloomberg video: electricity may become AI's biggest constraint
Switchyard Partners CEO Joe Kaiser tells Bloomberg Open Interest that power, not chips, could cap the AI boom, arguing data centers may increasingly relocate to stranded power. He also claims capitalism could produce better AI guardrails.
Bloomberg Technology·1 hour ago
Cohere releases North Small Translate model
Cohere·1 hour ago
AWS publishes guides for Amazon Quick Automate agentic workflows
Four AWS posts cover Amazon Quick Automate, a multi-agent automation capability in Amazon Quick Suite, including an end-to-end RFI questionnaire workflow and Outlook email automation. One guide cites a Gartner study finding AI saves sellers nearly five hours a week.
AWS AI Blog·1 hour ago
AWS details model-agnostic PII detection with LLMs on Bedrock
Configurable, instruction-driven detector runs on any LLM managed on Amazon Bedrock, evaluated on five public PII corpora across nine LLM-based detectors including OpenAI PrivacyFilter.
OpenAI spotlights blind brothers using ChatGPT to navigate life
Bradford and Bryan Manning, who live with Stargardt disease, use ChatGPT to turn visual information into guidance for daily tasks like navigating crowded streets. The brothers also run a nonprofit.
YouTube·1 hour ago