We use cookies to measure how the site is used and how our ads perform (Google Analytics and Ads, Microsoft Clarity, and the Meta, OpenAI and X ad pixels). “Essentials only” turns off advertising cookies everywhere, and analytics too in the EEA, UK and Switzerland. Privacy Policy
Tool Retrieval is in preview in the Gemini Interactions API: add a tool_search tool, set defer_loading on the long tail, and the server loads parameter schemas on demand. What that changes for tokens, prefix caching, MCP catalogs and debugging — and where the saving does not show up.
Job listings spotted by Bloomberg show Apple building an autonomous agent that browses, codes, controls apps and coordinates sub-agents — the last major platform holder to enter the computer-use race.
SpaceX Super Intelligence announced The X Subscription on Oct. 12, 2026, unifying Grok, Cursor and X Premium into tiered plans from $8 to $200 per month with a shared AI usage pool across chat, agents, API and coding workflows.
Anthropic's Frontier Red Team demonstrated that models under 30B parameters can set up inference on unfamiliar hardware, GLM-5.2 completed end-to-end replication across an emulated Android fleet, and the trend line puts a small model at the replication capability threshold by the end of 2026.
A Microsoft Research paper finds long-running agents get stuck in idea basins. Fork-and-flush - branch the agent, flush the chat, keep the workspace - reached a 0.78 min-max-normalized average score versus 0.47 for single runs, a 66 percent relative gain at equal compute across 13 tasks.
Epic has delayed its Marvel and Star Wars shooter indefinitely and told staff there will be no more hand-written code, with Claude-generated code and full AI art workflows. What the first big-studio AI-first mandate means for teams building on coding agents.
OpenCode 2.0 makes the client/server model the default: one background service shared by the TUI, web and desktop, Code Mode scripts for MCP tools, background subagents, 80-plus plugins and an embeddable SDK the launch note calls still in development, plus the opt-in upgrade path from 1.19.0.
Meta's Superintelligence Lab released Muse Wonder and Wonder Full, its first frontier-class models, claiming benchmark parity with Anthropic and OpenAI while emphasizing long-term active memory for coding agents.
Automater Lite 0.1.31 hardens AI tool install/update flows on Windows and macOS, verifies installed tool versions, preserves per-user desktop app locations, and repairs local voice that had broken in Linux AppImage builds.
Microsoft's Oct. 12 announcement says Maia 200 already serves Copilot workloads and that the Maia family will scale to multi-gigawatt deployments from 2027, when Maia 300 enters production. What is shipped, what is roadmap, and how to read the efficiency claims.
Ollama v0.40.4-rc0 is a pre-release that updates llama.cpp to b11521 for decision models and EmbeddingGemma 2, reuses manifests in inference routes, compacts MLX state, and surfaces local models in the Codex picker. Here is how to test it.
The marketing manifest at automater.ai/release.json now reads marketing-2026.10.12.4, built 2026-10-12T19:39:51.602Z, with both rollback fields at marketing-2026.10.12.3. What the record verifies, what it does not state, and the public News section around it.
GitHub mitigated Oct. 12 degradation in Actions, Issues and Projects but notifications remain backlogged, while Vercel confirms GitHub-connected builds are failing. What agent teams should pause and check.
Automater marketing build marketing-2026.10.12.3 built 2026-10-12 identifies current site identity, domains and rollback target; News index shows Subscribe via RSS, topic chips with Filter by topic and dated stories; editorial method documents AI-assisted two-reviewer workflow.
llama.cpp pushed three pre-release binary tags in under four hours on 11 October. The newest, b11558, fixes OpenCL image-limit handling for Q4_K dense bin kernels plus q4_0 broadcast and rms_norm subgroup issues. Here is what changed and how to roll it out safely.
Qwen Code's stable v0.25.1 moves multi-agent work onto sessions, removes the thread backend behind A2A, and lands with Desktop v0.25.1 and TypeScript SDK v0.1.19 bundling the same CLI. What operators should canary, pin and re-check.
OpenCode v1.19.0 stops installing updates automatically, adds /update and opencode upgrade --major as the explicit path to OpenCode 2, and validates PTY connection origins before existence checks. What pinned agent fleets should change now.
The Deno team is moving to Cloudflare to improve self-hosted Workers and Durable Objects. Existing runtime and hosting users face separate support deadlines while the combined platform remains a development plan.
Qwen Code v0.25.1-preview.2 enforces the per-model concurrency cap on foreground subagents, defers agent and goal declarations by default, and lands H4 child-agent, H5 channel and H6 automation record contracts plus a cross-node EventTransport design.
Google made Data Agent Kit generally available on October 9, 2026: a free collection of MCP tools and Google-authored agent skills connecting more than 15 Google Data Cloud services to VS Code, Antigravity, Cursor, Claude Code and Codex for schema inspection, query authoring and end-to-end pipelines.
v2.1.296 ships a `code` key in the Claude apps gateway's managed.policies[] that applies the same policy to Claude Desktop's Code tab and the CLI, plus autoCompactWindow and a workflow-subagent model pin for subagent fleets.
OpenAI's Codex CLI 0.162.0, published October 8 on GitHub, adds managed Git worktree tools for parallel agent lanes, Command Center task pinning, transcript copy shortcuts and clickable URLs, plus sandbox, Windows and provider fixes.