01Log
Archive
The full log, newest first.
CLM
Contrastive Language Models trains System One models via an objective that connects states and actions, serving CLM-8B with fine-tuning APIs for researchers.
ARTEX
ARTEX is an autonomous AI penetration testing system combining a Go backend with Next.js frontend, providing security professionals.
Hermes Agent 0.21.7
ā¢ Hermes Agent v0.21.7 is a patch release dated October 11, 2026. ā¢ This tag rolls up the ~204 PRs merged since v0.21.6 into a stable release for the desktop app (macOS Apple silicon, Windows), the Microsoft Store and MSIX, the installer, Docker and Hermes Cloud. ā¢ Updates now follow tagged releases. A source install of the official repository updates to the latest published vX.Y.Z release instead of every commit on main. An install already newer than the latest release stays where it is and waits for the next one; nothing is moved back. To keep following every commit: hermes update --set-chan
OpenClaw 2026.10.5-beta.2
ā¢ Updates finish and recover: openclaw update settles stale handoff leases, completed receipts, and legacy 2026.9.8 recovery records instead of stopping on update-recovery-pending, keeps the service heap limit and webhook listeners, handles Windows short-name paths, and no longer reports accepted handoffs or successful updates as failures. ā¢ Doctor keeps your plugins and data: Doctor preserves installed plugins' configuration when path safety checks block them, leaves malformed UTF-8 root memory files untouched, and recovery copies capture Canvas and Memory Core files before migrations move th
AGNO 3.1.3
ā¢ Version: 3.1.3. ā¢ Durable External Agent Sessions: ClaudeAgent sessions now resume on any replica. The Claude SDK session id is stored on the Agno session (#10903), and with PostgresDb, AsyncPostgresDb, SqliteDb, AsyncSqliteDb every transcript line is mirrored into the new agno_transcripts table. A stored session that cannot be resumed falls back to a fresh SDK session seeded with the Agno history. Claude and Codex non-streaming runs now keep tool calls and results in run history.ā¦
GenZ-LIO
Implements the IEEE T-FR 2026 paper GenZ-LIO for robust, computationally efficient LiDAR-inertial odometry across confined and open environments.
portal
Rust transport layer that closes the robot teleoperation and inference control loop over WebRTC through NAT without VPNs or forwarded ports.
Spotter
Official codebase for Spotter, which adds an intervention layer between a frozen embodied model and a VLM to improve manipulation tasks.
FlashDexRetarget
Official implementation of FlashDexRetarget, which accelerates dexterous manipulation data generation via multi-motion retargeting.
Calibrex
Command-line tool that checks whether a robot's sensor calibration remains valid from ROS bags with per-axis verdicts for LiDAR, IMU, camera, and GNSS.
OpenClaw 2026.10.2
ā¢ OpenClaw 2026.10.2: the first upgrade from stable 2026.10.1 now completes cleanly, successful in-place updates no longer end in errors, and Windows, cloned-filesystem, and private-Node update paths finish instead of stalling or rolling back. ā¢ Doctor no longer stops on residual legacy imports, stale Gateway locks, slow state inspection, memory reindex locks, or allowlists for removed channel accounts, and stale-receipt sessions keep their current data. ā¢ OpenAI and Anthropic API-key routes now default to provider-native compaction, and compaction recovers when the primary auth profile is coo
SWE-Duel: Towards a Self-Scaling Arena for Coding Agents
arXiv paper (submitted 8 Oct 2026) proposing SWE-Duel, a self-scaling arena for evaluating coding agents; authors Wei Minn, Zhensu Sun, Lwin Khin Shar, David Lo.
OMP 18.10.0
Raises the default context window for `openai-codex/gpt-6.1-sol` and `gpt-6.1-sol-wm` from 272K to 500K, matching OpenAI's new Codex default; extended context still opens 922K. Also adds `omp session pack`/`unpack`/`moved` to move a session with its artifacts to another machine (resuming the copy left behind forks it), `omp auth export`/`import` to carry model sign-ins, and `/fork` to create a new Tern pane while a turn is streaming.
Multics
Multics generates video from a still image, giving builders a self-hostable path to animating product shots, storyboards, or ad creative.
Muse Wonder
A frontier reasoning model built for agentic personal and professional work, taking text, images, video, audio, and PDFs through one endpoint.
H2O Lightning 4B
H2O Lightning 4B takes image-text input in a footprint small enough to co-locate with other services on one machine.
Gemini 4 Argon spotted in A/B testing on Antigravity
Reportedly reaching some users through A/B tests on Antigravity; early outputs shared on X look like a clear step up from Gemini 3.8 Flash. Spotted in testing ā not a confirmed launch.
The gap in shared understanding of LLM capability is widening
HN thread (186 points, 285 comments) on Karpathy's argument that the gap in shared understanding of LLM capability is widening.
manimgx
ManimGX is a GPU-accelerated animation engine that renders math and 3D videos using Manim's Python API.
codex-profiles 1.3.0
Adds a native profile and workspace menu with usage, a `usage` CLI command for remaining Codex quota, low-quota alerts, and a Raycast extension for profiles and workspaces. Maintainer waived Desktop smoke checks ā real Desktop behavior is unverified.
vsl
This V library enables AI and high-performance scientific computing with linear algebra and numerical differentiation tools.
gdl
This C++ compiler provides a free, GPL-2.0 licensed implementation of IDL for data analysis and visualization in astronomy and medical imaging.
bblean
This Python library provides memory-efficient BitBIRCH clustering for high-throughput analysis of massive molecular libraries.
WERM
This HTML project implements a 302-neuron network based on the real C.
VeriTMM
This Python workbench offers verifier-first transfer-matrix methods for multilayer optics with differentiable inverse design and machine-readable validation.
OpenClaw 2026.10.5-beta.1
ā¢ Safer updates: updates finish across devices, filesystems, Windows elevation, and interrupted activations, and recover from stale leases or changed evidence instead of stranding the install or blocking future updates. ā¢ Restart-safe conversations: interrupted turns, first messages during worktree preparation, and deferred tasks resume after a Gateway restart, and user input and committed session data survive restarts, concurrent writes, and failed publications. ā¢ Faster, steadier Gateway: session, transcript, history, and metadata reads move onto worker threads with far less SQLite work per
robot-heads
Provides animated 3D robot heads in React on 2D canvas showing agent states like thinking, useful now for dependency-free chatbot and AI interfaces.
Design System React
Implements Salesforce Lightning Design System in React 19 with TypeScript, SLDS 2, dark mode, and accessibility.
frosty
Creates a mobile Twitch client for iOS and Android supporting emotes from 7TV, BTTV, and FFZ, valuable now for enhanced Flutter-based streaming viewers.
Elementary UI
Enables SwiftUI-style declarative apps in the browser via WebAssembly with reactivity and tiny Embedded Swift binaries.
mygo
Builds desktop apps in Go using webview frontends like WKWebView or native GPU UI with generated TypeScript clients for single binary distribution.
OpenClaw plugin SDK moves SQLite runtime writes into workers
Plugin SDK refactor (#168857) moves SQLite runtime writes into workers ā a change to where plugin runtime writes are executed; worth retesting any OpenClaw plugin setup that uses SQLite.
Qwen Code 0.25.1
ā¢ Version 0.25.1. ā¢ replace selected remote Hosts without losing bindings. ā¢ wait 60s before the E2E build artifact download retry (#13284).
ohmygame
OhMyGame consolidates AI-powered game design, asset creation, coding, playtesting, and publishing into a single app.
Astro Loop
Astro Loop is a GPL-3.0 Android roguelike shooter with one-finger controls, autofire, no ads or tracking, and no internet permission.
MCP servers v1.0.0
server-everything, server-filesystem, server-memory, and server-sequential-thinking publish at 1.0.0 on npm; mcp-server-fetch, mcp-server-git, and mcp-server-time publish at 2026.10.10 on PyPI. The everything server also cleans up subscriptions on session disconnect.
500B Tokens Later: Letting AI Agents Decompile a First-Person Shooter
HN thread (140 points, 103 comments) on using AI agents to decompile a first-person shooter after 500B tokens.
OpenCode 1.19.0
ā¢ Version 1.19.0 ā¢ Run opencode upgrade --major to upgrade to OpenCode 2. ā¢ Allow the OpenAI ultrafast service tier.
JEV 35B
JEV 35B targets classification and decision tasks, so it can gate or route traffic without paying for a full chat model.
OpenClaw 2026.10.1
ā¢ 2026.10.1-beta.2 is a hotfix beta covering 40 intervening commits since 2026.10.1-beta.1. ā¢ Updates and Doctor: recover archive migrations and interrupted updates without leaving managed Gateways stopped, and restore Gateway lifecycle commands on macOS 12 and disabled Windows tasks. ā¢ Replies and messaging: preserve completed answers after silent follow-ups, let Discord corrections pass deferred messages, and restore Telegram sends and buttons after channel-routed handoffs.