Everything that mattered in AI, one page a week.

OpenAI said it temporarily slowed scaling, including a two-week pause in deployment training, while its largest planned frontier RL run remained on hold. Preliminary evals suggested an upcoming model could meet the Critical cyber threshold; monitoring was extended to all tool-using inference.

Activation classifiers at every sampled token escalate to automated investigators with a 30-minute alert target. OpenAI put monitoring overhead at roughly 20% of observed inference compute.

Claude ran 24–48-hour autonomous campaigns against 16 protein targets: researching targets, choosing epitopes, installing and running open design models, and making every design decision without human input. Two independent labs synthesized every design exactly as delivered.

354 of 1,320 designs bound — a 26.8% hit rate — across 14 of 15 interpretable targets. On RBX1, 28 of 90 designs bound at 3.9 nM versus 9 of 245 human competition entries at 45 nM.

OpenAI secured roughly 8 GW at the former Portsmouth Gaseous Diffusion Plant in Ohio, with SB Energy building and operating it under a 20-year lease. Nvidia’s filing disclosed credit support capped at $105B for the first 4.25 GW and a $1.5B equity investment in SB Energy.

The site is exclusively Nvidia compute; the first 800 MW is expected in 2028. The deal includes community funds and Codex credits for Ohio students.

Nous Research shipped Bot Mode for Hermes Desktop, replacing the single-agent session list with a roster of named bots. Each bot is a real Hermes profile with its own chat, memory, skills, pinned model and avatar, and they message each other through a persistent Agent Inbox and hand work off by mention. It ships as a desktop plugin rather than a core patch, MIT-licensed, and is now bundled on by default.

The announcement drew real traction — past 900,000 views on the company's own post within a day, and reposted well beyond its usual circle. Hermes Agent is already on this timeline twice this quarter: a harness comparison on 7 August where it posted the lowest cost per success at $0.46, and a $1.5B valuation round reported on 13 July.

xAI released Grok 4.6 with a 500K context window and a $2/$6 per-million-token API price. It improved on Terminal-Bench, APEX-Agents, CursorBench and DeepSWE after a longer supplemental training run using model-generated reasoning data and agentic RL.

It is live in Cursor, Grok Build, OpenRouter, Vercel and Cloudflare — a distribution advantage that matters as much as the score.

xAI opened Grok Bot in beta on 11 August for SuperGrok subscribers on desktop and iOS, with Windows and Linux builds out and Android promised. The framing is deliberately not a chatbot: each Bot gets its own cloud computer, signs into the tools and sites you already use, and carries a job to the end, returning only when something needs approval. The launch post's own line is that there is a huge difference between 90% done and 100% done.

It arrives through an unusual corporate route — a SpaceX rollup that folded xAI and Cursor into one company — and rides distribution rather than novelty: Grok was already reported at roughly 117 million monthly active users in March 2026.

Meta released Muse Glimmer under Apache 2.0, optimized for always-on local agents, function calling, coding and judge workflows. The three-phase recipe used teacher-logit distillation, longer-context agent data, SFT, on-policy distillation and RL.

Weights landed with developer docs and integrations for llama.cpp, MLX and ExecuTorch. A 30B agent model that runs locally changes the privacy and latency calculus for personal tooling.

Anthropic reported an unreleased Claude combining results from separate papers to raise the proven lower bound on zeta zeros on the critical line from 41.6% to 67.2%. Claude also produced a publicly checkable Lean 4 proof; it did not prove the Riemann hypothesis.

Secondary reporting described roughly 36 hours of autonomous work across 60 subagents and 650 failed ideas. The important artifact is not the number, it is the formalization.