Runway just launched Dev plugins for Claude Code, Codex, and Cursor, connecting coding agents directly to its generative media platform via MCP. Agents can check credit balances, pick models, and trigger video workflows right inside the editor without storing API keys. It basically turns your coding agent into an end-to-end multimedia builder.
xAI just introduced Data Bot in Grok Bot, an internal AI teammate that lets any employee query data warehouses and run deep analyses without waiting in line. Curated by data scientists with shared institutional memory, it can live in Slack to post updates and handled an estimated 70 hours of manual work in just one week.
Andrej Karpathy just built HackerTube, an alternative YouTube recommender, in about two hours using Claude Opus 5.5. Frustrated by YouTube's noisy recommendations, he used Claude Code to filter 43 million Hacker News comments down to 6,454 heavily recommended videos with no login required. Why it matters: It proves how frontier AI lets solo developers ship custom, algorithm-free tools in a single afternoon.
DeepSeek just open-sourced dsh-superagent-prototype, an experimental background agent designed to handle GitHub tasks without cluttering human attention. It runs tasks in isolated sessions and only pings developers when a pull request is ready or human intervention is strictly required. Keep in mind that DeepSeek frames this as research code rather than an official product.
DoorDash CEO Tony Xu just revealed that third-party AI agents have placed only "several hundred" orders in total, compared to the company's 3 million daily deliveries. Xu warned against being delusional about agentic commerce, though DoorDash is still experimenting with its own in-house bots.
Aura just open-sourced Autopilot, a real-time game playtesting agent for Unreal Engine 5, Unity 6, and Godot 4. Developers simply prompt it with an objective, and the agent reads project code, plays through levels, and visibly improves its skills over repeated runs while generating bug reports. It also connects directly to Claude Code and Codex via MCP.
Neuralink just upgraded its brain implant to hook directly into xAI’s Grok. An ALS patient can now mentally trigger an AI agent to take over the screen, type in iOS apps, and even place phone calls hands-free. Why it matters: Instead of painstakingly clicking cursors pixel by pixel, brains can now issue high-level intent directly to AI agents.
CMU developer Zheyuan Kong just launched Auto-Company to cure the endless meeting plague among multi-agent demos. Fourteen persona-driven agents brainstorm and debate, but a hardcoded convergence loop forces a hard GO or NO-GO decision, triggering Claude Code to write and deploy real code by round three. It runs locally via Bash and Markdown, but requires full system access.
Google just introduced Gemini 4 Carbon, its newest frontier model built specifically for long-horizon agentic tasks and cybersecurity defense. It grabbed the #1 spot on the Artificial Analysis Intelligence Index v5.0 and broke records in vulnerability detection and patching benchmarks. It costs $2 per million input tokens and $10 for output, rolling out first to select cyber defenders.
Researchers just introduced DecisionRAG, a new framework that ditches traditional vector retrievers in favor of decision models. Instead of slowly generating reasoning tokens, the model navigates documents like a library hierarchy, scoring entire branches in a single forward pass. It searches through 2.68 million Wikipedia passages in 0.7 to 1.8 seconds, beating dense retrievers with rerankers by double digits.
OpenAI just launched GPT-6.1 Terra, a faster and cheaper model built specifically for visual reasoning and desktop computer use. It cuts standard API pricing by 50% down to $1 per million input tokens, while running tasks over three times faster than GPT-6 Astra. Basically, automating routine office software and technical drawings just became practical at scale.
OpenAI briefly listed two new release-candidate models on its API before pulling them down minutes later, exposing a dual-VM setup built for autonomous agents. A benchmark on OpenAI's blog also revealed a mystery internal model hitting 87.5% on engineering tasks, basically signaling the next flagship is right around the corner.
Pagekin is a newly open-sourced AI novel-writing agent designed to prevent story drift and broken character lore. Instead of dumping entire books into context, it structures world-building, character files, and chapter blueprints to load context on demand, offering granular scene rewrites and revision diffs. The tool is free on GitHub, but users must bring their own model API keys.
Kimi just launched K3.1 Pro, a 3T-class open-weight MoE model packing 2.8T total and 104B active parameters with a 1M-token context window. It handles full-project game coding across Unreal, Unity, and Godot, while cutting finance hallucinations by 65%. The open weights are slated for release on Nov 4, 2026.
Anthropic just rolled out its unified Claude experience to Enterprise customers across web, desktop, and mobile. Instead of picking between chat and Cowork beforehand, Claude now answers quick queries normally and automatically spins up isolated cloud VMs for complex coding and document tasks that keep running after you close your laptop. Admins also get a new Efficient mode to keep costs in check.
OpenAI just made its Decisions API generally available, using GPT-6 Luna to handle classification and routing decisions up to 10x faster than standard calls. It accepts text, images, or native JSON across up to 400 questions per request. Pricing is $0.10 per million input tokens with zero output fees.
Microsoft just unveiled MAI-DxO in Science, an AI diagnostic system that acts like a panel of virtual physicians debating complex cases across rounds. On tough New England Journal of Medicine cases, it hit 83% diagnostic accuracy compared to 36% for doctors using reference materials—all while cutting estimated test costs by more than half. Why it matters: instead of answering one-off multiple-choice questions, the AI learns to order the right sequential tests, moving clinical superintelligence closer to real-world hospital workflows.
Anthropic just donated Agent Skills—its open standard for teaching AI agents specialized tasks—to the Linux Foundation's Agentic AI Foundation. Already adopted by over 70 platforms including ChatGPT, Cursor, and VS Code, the standard boasts nearly 280 million installs. Engineers from Anthropic, Google, and OpenAI will now maintain it independently, keeping agent workflows truly vendor-neutral.
Anthropic just launched its fully managed Claude Platform on Google Cloud, giving enterprises same-day access to native API features like prompt caching, Managed Agents, and computer use. Teams can stick with their existing Google Cloud IAM permissions and consolidate billing into a single invoice. It basically eliminates the classic trade-off between using cutting-edge models and staying within corporate cloud guardrails.
OpenAI just announced a partnership with Google Cloud to bring its frontier models and Codex workflows into Model Garden. GCP customers can soon purchase OpenAI models via Marketplace and burn down existing cloud spend commitments, with integrations into Google Antigravity and Gemini Agent coming shortly. Why it matters: Even as Microsoft remains OpenAI's primary cloud backer, the deal gives enterprise devs freedom to run OpenAI models directly inside Google's infrastructure.
