Latest news, beside the leaderboard

On Oct 9 ClaudeDevs said every Pro and Max user on the Claude Code Projects waitlist has been let in. A project is a long-running coordinator conversation that spins each task into its own thread, usually a parallel cloud session on its own branch that opens a PR when needed; threads can also run locally via Remote Control. Still public beta; not yet on Team or Enterprise.

Long-horizon coding agents tackling complex software repositories depend on timely guidance, yet conventional LLM critics often degrade agent outcomes by misinterpreting in-progress exploratory commands or offering ephemeral advice that is never tracked. Researchers from Rutgers and Lehigh University unveil Opera (arXiv:2609.33987), an open-source verbal critic framework that treats diagnostic corrections as persistent notes tracked until root problems are definitively resolved. Opera orchestrates reviews through hybrid periodic and event-driven triggers, verifies proposed critiques against visible terminal artifacts, and monitors downstream trajectories to distinguish mere superficial compliance from genuine resolution. As a test-time intervention, Opera elevates baseline agent resolve rates by 12.4 percentage points on Terminal-Bench 2.1, 15.0 on SWE-Bench Pro, and 8.9 on DeepSWE v1.1 across four foundation backbones. Furthermore, fine-tuning Qwen3.5-9B on Opera-guided rollouts yields a 10.2 percentage point gain on held-out repositories without any test-time critic, preserving stability across heterogeneous execution harnesses (such as OpenHands to Terminus-2).

Endowing artificial agents with general intelligence requires inferring environment dynamics and refining world models continuously in the wild. Fine-tuning models directly is prohibitively expensive and precipitates catastrophic forgetting. Researchers from University College London (UCL) and Huawei Noah's Ark Lab UK present Memento 3 (arXiv:2610.11794), a framework enabling frozen LLM agents to achieve model-based Recursive Self-Improvement (RSI) purely via external memory. The agent maintains a natural-language rulebook representing its evolving hypotheses of environment physics, which is compiled into executable code for fast simulation and action planning. Governed by a continuous observation-reflection-revision-compilation-verification loop, the system leverages prediction errors to refine rules. Edits are merged only after passing LLM semantic alignment audits and deterministic cell-exact replay verification. On the rigorous ARC-AGI-3 abstraction benchmark, Memento 3 solves 100% of levels across all 25 public environments, scoring a perfect 100.0 Relative Human Action Efficiency (RHAE) while utilizing only 44% of human action counts. In Atari Pong, a synthesized feedback controller shuts out opponents 21:0 across three evaluation seeds with zero test-time LLM inference.

Modern AI coding agents such as Claude Code, Codex, and terminal assistants routinely execute agent skills—instructions and executable scripts bundled in SKILL.md specifications—with full local user privileges. Developers share these skills primarily by copying files across repositories, inadvertently forging a shadow software supply chain without centralized registries, semver tagging, or provenance tracking. A seminal empirical audit by Fahd Seddik (arXiv:2610.11169) introduces Skill Constellations, constructing the first dated copy network of agent skills derived from git provenance across 2,193,119 skill adoptions on GitHub. The investigation demonstrates that a minuscule set of root repositories generates nearly all downstream copies, yet GitHub star counts correlate poorly with true lineage sources. Critically, downstream copies almost never adopt upstream security patches. By fitting a predictive copy network model, auditing just the top 100 identified root repositories halts 14.9% of subsequent high-risk skill adoptions (compared to a negligible 0.5% when inspecting the top 100 starred repos), proving the urgent necessity for decentralized package ecosystems to adopt versioned references over loose copies.

On 2026-10-09 developer t4t5 open-sourced omdrop-owl, a radio backend for the Omarchy plugin omdrop that gives two-way AirDrop with an iPhone on the Framework 13's MediaTek MT7925 while Wi-Fi stays connected. It builds on Seemoo Lab's 2019 OWL/OpenDrop reverse engineering rather than a from-scratch effort. README self-tests: ~1.3 MB/s receive, ~490 kB/s send; the AP must be on channel 6/44/149.

37signals opened its hosted MCP endpoint at https://mcp.basecamp.com/mcp: OAuth sign-in, read-only or full access, free on every Basecamp account, covering projects, to-dos, card tables, docs, calendar and Campfire.

OpenAI's Oct 9 dots update lets you create and customize a dot from the ChatGPT iOS and Android apps. Dots can now start work in Codex, follow up on existing threads using ChatGPT conversations, Codex threads and automations as context, and read and edit ChatGPT Work scheduled tasks.

On Oct 9 OpenAI opened composer predictions in beta in the Codex desktop app for personal ChatGPT Pro users. After each response Codex may prefill your next message from the current thread; press Tab to accept and edit before sending. Local and SSH threads only, with GPT-6 Astra and GPT-6.1 Sol; generating predictions does not count toward usage limits during the beta.

SpaceXAI's official @bot account announced on Oct 9 that Grok Bot is rolling out its own email inboxes. Ask your Bot to claim one and it gets a @mail.grokbot.com address it can use to sign up for services, contact businesses or schedule time; team admins must enable it.

Anthropic expanded dynamic workflows in Claude Managed Agents to public beta. With workflows enabled in an agent's multiagent block, Claude writes a workflow program that runs many agents in background phases and combines their results, aimed at audits, migrations, deep research and cross-checking.

On Oct 9 Pine AI released Pine Computer in private beta: cloud virtual computers plus a harness and runtime built for AI, driven by an API across browser, files and shell. Instead of repeated screenshots, the OS, browser and apps push change notifications to the model ("epoll for AI perception"). On SaaS-Bench v1.1 (106 tasks, 23 apps) GPT-5.6 Luna on Pine scored 78.3% on checkpoints vs 74.3% for Opus 5 + Claude Code, though its fully-resolved rate (27.4%) trails the latter (31.1%).

On Oct 9 OpenAI Developers announced a new Codex sandbox mode on Windows built on Microsoft Execution Containers (MXC), promising faster setup, stronger network enforcement and granular file access controls on compatible Windows 11 devices. Docs now list MXC as the recommended implementation, with elevated/unelevated as legacy fallbacks. It ships in Codex CLI 0.162.0, where features.prefer_mxc is off by default in the standalone CLI; the desktop app enables it via rollout.

Sonnet 5.5 cache hits now cost 0.05x base input ($0.10/MTok), which Anthropic says makes most agentic work ~20% cheaper. Max 5x/20x plans now include $100/$200 in monthly Claude API credits (Team $20–$100 per seat) usable on the API, Managed Agents and Agent SDK, alongside an official reference implementation for scheduled Managed Agents automations.

Nullframe (the Moda team, YC F26) shipped Voyager, a local desktop agent harness that drives creative apps through their own scripting APIs, ships free Design/Video/DAW engines, and works with your own Claude/Codex sign-in or API keys. The launch post passed 1.48M impressions.

On Oct 8 Odyssey released Odyssey-3, a foundation world model built as an autoregressive diffusion transformer. Odyssey-3 Pro (720p) scores 66.1 on Physics-IQ Verified video-to-video, the highest reported score, and 54.7 image-to-video. A free browser research preview (Flash variant) offers first/third-person navigation and free camera; API access is by request, with no open weights or public pricing. Odyssey says tens of hours of demos are enough to control various robot arms.

IQuest Research’s IQuest-Q1 (~320B total / 15B active, 524K context) targets repo-level coding, terminal work, and long-horizon tool use. Weights and serving images are on Hugging Face / GitHub. Self-reported same-pipeline scores: DeepSWE v1.1 64.6, NL2Repo 63.0, CyberGym 84.5, Terminal-Bench 2.1 83.2. Ships with SGLang/vLLM images and Claude Code / Codex integration notes.

Cline moved its open coding agent into the browser: connect GitHub, pick a repo and model, and each task runs in an isolated cloud sandbox that commits to a `cline/` branch and can open a PR—even with your laptop closed. Limits: 10 concurrent sessions and 50 new sessions/day. Providers: ClineFree, ClinePass ($9.99), and usage-billing. Cloud sessions do not yet load Rules, Skills, MCP, or Connectors.

On Oct 8 JetBrains released Mellum2.1 Thinking: same 12B MoE / 2.5B active / 131K / Apache 2.0 architecture, with gains almost entirely from large-scale RL in real sandboxes. Self-reported same-pipeline scores: SWE-bench Verified 47.0 (vs Mellum2 2.0), LiveCodeBench v6 82.0. On Hugging Face; vLLM serve now; GGUF/MTP coming soon.

On Oct 8 Anthropic launched the Cyber Mission: (1) free opt-in OSS Scanner for critical open-source projects using strongest models including Claude Mythos, with PoC/explanation/candidate patches and >90% expected true-positive rate, no human triage; (2) Critical Infrastructure Defense Program with 11 founding partners including CrowdStrike, Palo Alto, Dragos, and Rockwell. Project Glasswing has surfaced 29,000+ candidate vulns.

On 2026-10-08 xAI shipped grok-imagine-video-1.5-lite in the Grok Imagine API: text/image-to-video with native lip-synced audio, 1–15 s, at $0.02/s (480p), $0.03/s (720p), and $0.14/s (1080p, upscaled from 720p). No reference images, voices, or keyframes; those stay on Video 1.5 ($0.08/s). Batch API supported; no quality evals published.