Is Day 17 gone? Did Tibo fall asleep and forget?
Model intelligence feed
Model intelligence.
Follow the changes. Understand what comes next.
Editor’s pickOther providers
Valve's Answer to AI Slop and Clone Games on Steam: We Won't Police Developers' Tools
At Steam Indie Connect in Korea, Valve staff addressed AI slop and clone games head-on: no gatekeeping on AI tools, and the best defense against clones is community and steady updates.
Latest stories337
Newest published133134135136137138139140141142143144
DetectorScope Explains Its AI-Text Calls. Why That Still Isn't Proof of Cheating
DetectorScope is an open Google DeepMind and University of Pennsylvania research project that rebuilds a neural AI-text detector so its score is a sum of named, validated features that can be read and located in the text. The paper was submitted to arXiv on October 21, 2026; GitHub News on October 25 says it will appear at EMNLP 2026. The README says this is not an official Google product, that detectors err on human-written text, and that they must not be used for academic integrity, hiring, or publication decisions. We have not run the demo ourselves.
DetectorScopeRead story
Sweden's Government Offices Block ChatGPT and Claude Sites on Staff PCs—Not a Nationwide Ban
Sweden's Government Offices (Regeringskansliet) will block the ChatGPT and Claude websites on staff PCs from Monday, October 26, 2026, according to Sveriges Radio Ekot and SVT. The move covers browser access on office computers—not a nationwide ban, and not a removal of Microsoft Copilot. Staff can still use ChatGPT and Claude models inside the offices' Intric AI assistant platform. DeepSeek was restricted earlier at the same agency.
ChatGPT／ClaudeRead story
Same Copilot Loop, More Agents: Agensh Claims Pass Rates Rise Toward 1,024 Workers
Microsoft Research open-sourced Agensh (MIT), a decentralized multi-agent coding harness with no central orchestrator. README News marks 2026-10-24 OSS; GitHub release v1.0 published about 16:03 Beijing time on 2026-10-23; the arXiv preprint is dated 2026-09-22. On the five hardest ProgramBench tasks with GPT-5.6-sol (high) and a six-hour budget, the authors report mean final test-pass rising from 19.31% with 1 agent to 28.78% with 128 agents, and on Pandoc from 33.89% with 1 agent to 55.06% with 1,024. Figures are from the paper and README—not our retest; for reference only.
AgenshRead story
Same Open Model, Different API: Why Speed and Cost Both Change
Vals AI’s Inference Provider Index (page Updated 10/20/2026; Updates dated 10/22/2026) runs the same open-weight models—DeepSeek V4.1 Flash, Kimi K3, and MiniMax M3—on first-party APIs and hosts such as Together AI, Fireworks, and Baseten. Across 165 agent tasks, overall accuracy stayed within error of the first-party API. Cost per task rose by as much as about 2.3× by host, and for all three models the creator’s API was neither the cheapest nor the fastest. Figures below are from Vals AI, for reference only.
Inference Provider IndexRead story
Claude Max and Team Monthly Platform API Credits ($100 / $200)
Claude Max and Team plans include monthly credits. Max 5x gets $100 a month and Max 20x gets $200; Team pools $20 per Standard seat and $100 per Premium seat into one balance capped at $500. Free, Pro, and Enterprise are not eligible. After seven days on an eligible plan, link one Console organization under Billing to claim. Credits cover the Claude API, Playground, Managed Agents, and Agent SDK—not interactive Claude Code or app overages. They refresh each billing cycle, do not roll over, and stay separate from Claude / Claude Code plan limits. We have not tested claiming; for reference only.
Claude Max / Claude Team / Platform API creditsRead story
Aokigahara Infostealer Targets Claude, Codex, Cursor and Copilot Credentials
On October 23, 2026, Check Point Research published an analysis of Aokigahara, a newly observed infostealer (earlier versions may have been known as Kamidere or Stygian). Beyond browsers, wallets and other classic targets, its aisessions handler collects authentication files, session data and configs tied to Claude, ChatGPT Desktop, Codex, Cursor, Windsurf, GitHub Copilot, Gemini CLI and similar tools. An ide handler also reaches Git, SSH and cloud credentials. Stolen AI credentials could enable misuse or resale; the analysis does not confirm how operators monetize the data. We have not re
Aokigahara / Kamidere / Stygian (infostealer)Read story
Anthony Morris Leaves Anthropic: Claude Code Desktop Author
On October 24, 2026, Anthony Morris (@amorriscode) posted that yesterday was his last day at Anthropic. He wrote that he created Claude Code on desktop because he believes everyone can be a builder, thanked the community for feedback, and said he would share what comes next soon. The note is a personal farewell—not an Anthropic company announcement. Days earlier, Thariq Shihipar also posted about leaving Anthropic after work on Claude Code.
Claude Code (desktop)Read story
Chroma Joins Replit: Cloud Writes Stop Nov 22, Reads Dec 22
Chroma announced on October 23, 2026 that it signed a definitive agreement to join Replit and will wind down the existing Chroma Cloud service over about 60 days. Writes are disabled November 22, 2026 (read-only for 30 days); reads stop December 22, 2026. Open-source ChromaDB remains under Apache 2.0, with more private-repo code committed and Replit set to keep contributing. The wind-down applies only to hosted Chroma Cloud—not a shutdown of the open-source project.
Chroma Cloud / ChromaDBRead story
Gemini Can Export Interactive Apps to Google Sites
Google’s October 19, 2026 Workspace Updates note says you can export interactive apps built in the Gemini app to Google Sites. Export creates a new private site draft that keeps the interactivity; you then edit, publish, and share like any other Site. Available now on web, Android, and iOS for Workspace, Workspace Individual, and personal accounts with Gemini access—if Sites is on and you can create／edit. The post does not promise two-way auto-sync with Gemini after export.
Gemini → Google Sites ExportRead story
OpenAI Habitat: Storage Reads ~12x Faster — Not ChatGPT Answers
OpenAI’s October 23, 2026 engineering post (Part II) says Habitat’s global read p50 fell from roughly 120 ms to roughly 10 ms—about 12×—after placement near users where capacity allows, broad replication of small latency-sensitive data, Valkey lookaside cache with CDC invalidation, and a Rust service rewrite. That figure is storage-read median latency for Habitat, not model inference, full ChatGPT answers, every request, or tail latency. A custom exabyte-scale online store (RocksDB Cloud plus Raft) is still under construction and does not replace Cosmos／Habitat yet.
HabitatRead story
IQuest Opens Q1: 320B/15B MoE for Claude Code and Codex
IQuest Research has released IQuest-Q1, an open sparse Mixture-of-Experts model with about 320B total parameters and about 15B activated per token, aimed at agentic CLI and coding work. Context length is 524,288 tokens (512K). The website and Hugging Face card show SGLang and vLLM serving paths, plus examples for Claude Code and Codex CLI. Benchmark numbers are the project’s reported scores—not our measurements—and the card calls out text-only input and early-stage limits.
IQuest-Q1Read story
GitHub Copilot Usage Metrics Now Count OpenCode and Other Clients
GitHub’s Changelog says the Copilot usage metrics API now reports model usage from tools outside the IDE, including OpenCode. A new totals_by_client array appears in enterprise, organization, enterprise-user, and organization-user 1-day and 28-day reports. Server-side request counts close the gap left when those tools do not send editor telemetry. This is a governance and visibility update—not the same story as Copilot turning Teams notifications into fixes and PRs.
GitHub Copilot usage metrics / totals_by_clientRead story
From the Webmaster
Personal experiences, thoughts and notes.