https://sysnote.net/news/nolan-lawson-benchmarks-12-javascript-frameworks-with-opus-5-5-angular-and-svelt sysnoteen 2026-11-01T21:56:13+00:00 Nolan Lawson benchmarks 12 JavaScript frameworks with Opus 5.5: Angular and SvelteKit cost most, and framework MCPs often go unused https://sysnote.net/news/deepseek-harness-super-agent-design-router-workers sysnoteen 2026-11-01T21:55:37+00:00 DeepSeek Harness's "Super-agent" design: a Router with no Bash, and a single-session agent that never delegated https://sysnote.net/news/the-alleged-anthropic-numbat-checkpoint-on-hugging-face-doesn-t-look-trained-eve sysnoteen 2026-11-01T17:17:32+00:00 The alleged Anthropic "numbat" checkpoint on Hugging Face doesn't look trained: every norm layer is the same bell curve https://sysnote.net/news/gvs5h-s-qwen3-8-27b-vs-claude-fable-5-claim-checked-twice-the-grading-holds-the sysnoteen 2026-11-01T14:05:59+00:00 GVS5H's Qwen3.8-27B vs Claude Fable 5 claim, checked twice: the grading holds, the 2-point lead doesn't https://sysnote.net/news/balestriero-and-lecun-turn-the-platonic-representation-hypothesis-into-a-theorem sysnoteen 2026-11-01T13:41:52+00:00 Balestriero and LeCun turn the Platonic Representation Hypothesis into a theorem for JEPAs, and a 109M text model out-aligns 72B LLMs, with a captioner in the loop https://sysnote.net/news/qwenpu-puts-karpathy-s-stories260k-inside-qwen-3-5-running-on-a-vector-unit-comp sysnoteen 2026-11-01T12:47:38+00:00 QwenPU puts Karpathy's stories260K inside Qwen 3.5, running on a vector unit compiled into its weights https://sysnote.net/news/first-of-two-banked-codex-resets-lands-on-november-1-tibo-confirms-with-banked-1 sysnoteen 2026-11-01T01:18:52+00:00 First of two banked Codex resets lands on November 1, Tibo confirms with "Banked 1/2" https://sysnote.net/news/anthropic-confirms-claude-code-s-rule-against-recreating-existing-games-it-arriv sysnoteen 2026-11-01T01:18:44+00:00 Anthropic confirms Claude Code's rule against recreating existing games; it arrived through a remote prompt slot, tengu_heron_brook https://sysnote.net/news/openai-confirms-chatgpt-and-codex-will-share-one-token-based-usage-limit-from-de sysnoteen 2026-10-31T21:43:38+00:00 OpenAI confirms ChatGPT and Codex will share one token-based usage limit from December; Instant chat stays free of it https://sysnote.net/news/volta-infostealer-asks-claude-opus-4-6-to-rank-victims-by-what-they-re-worth-and sysnoteen 2026-10-31T18:30:13+00:00 VOLTA infostealer asks Claude Opus 4.6 to rank victims by what they're worth, and falls back to Kimi K2 if it refuses https://sysnote.net/news/shared-chatgpt-link-account-takeover-mathlive sysnoteen 2026-10-31T17:20:30+00:00 A shared ChatGPT link could take over accounts via MathLive's virtual keyboard; OpenAI paid a $5,000 bounty https://sysnote.net/news/openai-s-tibo-says-codex-usage-limits-get-two-more-resets-on-november-1-and-2-an sysnoteen 2026-10-31T12:29:57+00:00 OpenAI's Tibo says Codex usage limits get banked resets on November 1 and 2 https://sysnote.net/news/google-s-sleep-framework-lets-llms-consolidate-memories-offline-but-its-gpt-4-wi sysnoteen 2026-10-31T12:29:13+00:00 Google's Sleep framework lets LLMs consolidate memories offline, but its GPT-4 win compares fine-tuned and zero-shot models https://sysnote.net/news/seven-official-ollama-tags-of-qwen2-5-coder-3b-phi-3-5-and-llama-3-2-1b-can-t-wr sysnoteen 2026-10-31T05:22:47+00:00 Seven official Ollama tags of Qwen2.5-Coder 3B, Phi-3.5 and Llama 3.2 1B can't write code, and the cause is missing calibration https://sysnote.net/news/anthropic-claude-agents-wrote-harmless-looking-training-data-that-made-models-mi sysnoteen 2026-10-31T05:22:41+00:00 Anthropic: Claude agents wrote harmless-looking training data that made models misaligned and passed its own monitor https://sysnote.net/news/what-the-reports-behind-anthropic-s-drug-development-claims-actually-show sysnoteen 2026-10-31T01:17:51+00:00 What the reports behind Anthropic's drug-development claims actually show https://sysnote.net/news/openai-quietly-ships-gpt-6-rosalind-for-approved-life-sciences-labs-at-15-75-per sysnoteen 2026-10-31T01:17:30+00:00 OpenAI quietly ships GPT-6 Rosalind for approved life-sciences labs at $15/$75 per million tokens