AWS launches the open-source Strands AI agent harness and CLI, claiming it cuts costs 26% across seven benchmarks—and 77% versus Claude Code in Fable 5—while supporting Bedrock, Anthropic, OpenAI, Google, and local Ollama models without requiring AWS infrastructure.
Cognition adds Devin Cloud control to its CLI, allowing developers to hand off local tasks to persistent cloud VMs and access them with full SSH via devin ssh for editing, port forwarding, and scp transfers.
Xiaomi unveils MiMo-V2.6, an omni-modal AI model family it says significantly improves coding, design and cybersecurity through groupwise reinforcement learning, with its Pro version reaching 72.6 average@3 on DeepSWE v1.1 after the team freezes an unstable Mixture-of-Experts router.
Google launches Googlebooks, Android-based laptops starting at $899 that retain ChromeOS while centering Gemini features such as Magic Pointer, Rambler voice-to-text and Create My Widget; HP, Dell, Lenovo, Acer and ASUS models are available for U.S. preorder ahead of October 4 availability.
AMD launches the $3,999 Ryzen AI Halo local AI platform, pairing a Ryzen AI Max+ 395 chip with 128GB unified memory to run models up to 200 billion parameters and claiming 4% to 14% faster LLM token generation than Nvidia’s DGX Spark.
Google releases AX, an open-source Kubernetes-style orchestrator that runs autonomous AI agents in sandboxes across clusters using Task, Workspace, Gateway and Model primitives, while warning that its still-evolving protocols may undergo major breaking changes before a stable release.
Meta launches the SAM 3.1 API, letting developers use text prompts to detect, segment and track objects across video while preserving identity, with no model hosting or fine-tuning required and pricing starting at $2.50 per 1,000 images.
Vercel Labs releases json-render v0.21.0, adding experimental composition APIs and a playground model option while constraining AI-generated interfaces to developer-defined component catalogs, schemas, and actions.
mizorewww launches Laya-MLX, a native Apple Silicon runtime for Convai Innovations’ typed decision models that runs without text generation, PyTorch, or cloud APIs, reporting 7.39 ms median latency for its 322M multilingual model on an M3 Max.
Anthropic adds AGENTS.md support in Claude Code 2.1.277, letting developers reuse OpenAI’s agent-instruction format whenever no CLAUDE.md file is present and toggle the behavior through /config.