01Log
Archive
The full log, newest first.
Claude Code in the cloud: a field guide to cloud sessions
Addy Osmani's field guide to running Claude Code on its own machine: what changes, which workflows pay off, and how to connect GitHub on the first try.
Introducing Mistral Large 4: Le chonk
Mistral ships a preview of Mistral Large 4 — a 1 trillion parameter, 49 billion active parameter model trained on their own cluster of 3,800 NVIDIA Grace Blackwell GPUs.
Codex 0.161.0
Codex 0.161.0, tagged rust-v0.161.0 in openai/codex and published 2026-10-06 — release notes at the link.
Update your IDE to restore agent activity in Copilot usage metrics
Agent activity has been missing from Copilot usage metrics — GitHub's October 6, 2026 changelog post says updating the IDE restores it, and covers what to expect in your reports. Anyone tracking Copilot agent usage in org dashboards is undercounted until the client updates.
Gemini CLI 0.63.0
Gemini CLI 0.63.0 — bounds tool output size and optimizes memory lifecycle in long-running agent loops; fixes an infinite auth loop from file contention, headless keyring, and supervisor state drops (#28341); distinguishes missing MCP enablement config from malformed JSON; cleans up temporary directories when background shell execution exits.
OpenTPU – An open-source AI accelerator, developed by AI
HN front page at 210 points / 275 comments on an open-source AI accelerator from github.com/fesens; top comment asks why the labs aren't burning their frontier models into chips already.
Sharing AI Progress in Mathematics
OpenAI's post on AI progress in mathematics, 244 points / 190 comments on HN; links an openai/math repo whose preprints include a proof of Barnette's Conjecture.
NVIDIA CorrDiff Cosmo ERA5
NVIDIA's CorrDiff line lands on Hugging Face for teams doing atmospheric downscaling work who want the model weights in their own pipeline.
d1-omni 600M
A compact image-text model for local multimodal understanding on constrained hardware.
Interfaze 1 Lite
An image-to-image model for lightweight visual editing and generation pipelines.
d1-3B
A small multimodal model for local deployment and quick experimentation.
Claude Haiku 5.5
Built for high-volume summarization, subagents, and browser automation where throughput matters more than peak reasoning.
Cohere Transcribe 03-2026
Cohere's speech recognition entry gives builders a self-hostable transcription model to drop into their own audio pipeline instead of renting one per minute.
SkillForge: Co-Evolving Skills and Agents via Dynamic Skill Lifecycles
Co-evolves agent skills and the agents that use them through dynamic skill lifecycles, with memory-augmented RL strengthening LLM agents' abilities.
Code scanning AI Scan enablement status in security overview
Organization and enterprise administrators can now see AI Scan for pull requests enablement status in security overview. Filed under Improvement in the October 6, 2026 changelog.
ASCENT: Online Test-Time Training of Long-Horizon Agents via Self-Distillation of Verified Experience
ASCENT proposes online test-time training for long-horizon agents via self-distillation of verified experience; the page excerpt says it lets an LLM agent keep learning.
SearchJev: A Fast and Calibrated System-1 Model for Search Agents
SearchJev is presented as a fast, calibrated System-1 model for search agents.
Beyond Semantic Similarity: Performance and Costs of Agentic Retrieval for Complex Tasks
Compares the performance and costs of agentic retrieval against semantic similarity for complex tasks.
Advancing computer use with Ironclad
OpenAI post on advancing computer use with Ironclad.
Introducing Mistral Large 4
Mistral announces Mistral Large 4.
WatermelonDB React Database
WatermelonDB is a reactive, offline-first database for React and React Native apps that stays fast with tens of thousands of records. It uses lazy loading and observables to optimize client-side persistence across mobile and web platforms.
Adminer PHP Database
Adminer is a full-featured database management tool distributed as a single PHP file ready to deploy. It supports MySQL, PostgreSQL, SQLite, MS SQL, and other databases with a lightweight footprint.
Apache IoTDB
Apache IoTDB is a time-series database purpose-built for IoT workloads with a high-performance storage engine. It is an Apache top-level project specializing in efficient ingestion and management of massive device-generated data.
Quivr Content Engine
Quivr is an open-source engine that ingests continuous content streams, makes them searchable within seconds, and supports topic monitoring with alerts. It features durable ingestion, hybrid search, and a plugin system for formats, models, and business rules.
Goose Database Migrations
Goose is a database migration tool for Go that manages schema changes via SQL scripts or Go functions, working as both CLI and library. It supports multiple databases including Postgres, MySQL, and Spanner.
Ibis Python Dataframe
Ibis is a portable Python dataframe library that lets users compose lazy expressions (using DuckDB by default) and run them unchanged across 20+ database backends. It enables local iteration and remote deployment by changing a single line of code.
Flyway Database Migrations
Flyway is a database migration tool by Redgate supporting versioned schema changes across 50+ database systems. It is an industry-standard Java-based solution with broad enterprise adoption.
Self-Generated Feedback Destabilizes Test-Time Training: A Causal Decomposition of Long-Horizon Adaptation
HF Daily Papers: causal decomposition finds self-generated feedback destabilizes test-time training in long-horizon adaptation.
Mistral Large 4
Mistral Large 4 discussion — 1037 points / 680 comments on HN; model card at docs.mistral.ai/models/mistral-large-4-0.
An AI agent emailed researchers for help. It told us why
science.org report on ColonistOne, an agent given the task of 'telling more humans about its work' — 50 points / 80 comments on HN (flagged).
AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model
Introduces AdvSim2Real, a training framework for web agents that defends against adaptive prompt injection by simulating attacks in a web world model.
Kiro — Workflows, Safer Untrusted Workspaces, and Enterprise Sign-In Controls
Claude Sonnet 5.5 is now available with experimental support across the Kiro IDE, CLI, Crew, and Web for Pro, Pro+, Pro Max, and Power customers in us-east-1 (IAD) and eu-central-1 (FRA); the Oct 2, 2026 changelog update also ships Workflows, safer untrusted workspaces, and enterprise sign-in controls.
Devin: October 5, 2026
Session workspace gets a Terminal tab for running commands on Devin's machine — the Files tab can now be opened while Devin is asleep, and opening Files or Terminal wakes Devin. Also in this cut: a ⌘K command palette, Devin Review effort levels, and Slack user-group mentions rendering as named @group mentions in the webapp.
KunQuant Compiler
KunQuant is a compiler and executor that transforms financial factor expressions into optimized code, originally built for Alpha101 and Alpha158. Notable for its specialized optimization approach including AVX512 support, it targets quant researchers needing efficient batch factor computation.
Program Garden
ProgramGarden is a no-code node-based platform for building automated trading strategies, supporting Korean stocks and overseas futures via LS Securities. The AGPL-licensed project targets retail algorithmic traders in Korean markets.
Intro to Quant
Introduction-to-Quantitative-Finance is a Chinese open-source knowledge base combining multi-factor investment frameworks, AI+finance paper tracking, and curated learning resources. It serves students and researchers seeking structured quantitative finance education with academic depth.
Libitofin
libitofin is a Rust port of QuantLib offering pricing, risk, and numerical methods with Python and Go bindings. It provides a memory-safe, high-performance alternative to the C++ original, though APIs remain unstable pre-1.0.
Fauxnix Bash Translator
A deterministic bash-to-PowerShell translation layer built for AI agents that runs natively on Windows without WSL or VMs. It converts Linux-style commands with GNU-style output and error handling, solving agent-UX pain for Windows-based development.
Coucou Notch Companion
A Mac notch companion and iPhone app that monitors AI coding agents like Claude Code and Codex, allowing permission approvals from the Lock Screen or Dynamic Island. It provides a cross-device interface for watching agent sessions and approving permissions without leaving your current context.
OpenMausBot
An open-source Grok Bot alternative that runs teams of AI bots locally, each with its own virtual machine, personality, and connected apps. Users interact with them like contacts in a chat interface while approving their actions.