Topic · 5 stories
Industry
Stories
Theo drafts the Claude Code rate-limit announcement he wishes Anthropic had written (@theo)
X post · Story page
Quoting the announcement that Claude Code's weekly limits rise permanently by 25 percent for Pro, Max, Team, and seat-based Enterprise plans from September 14, Theo does the arithmetic left implicit: with the 50 percent promotion expiring, the permanent 25 nets out to a 17 percent decrease from today's limits.
OpenAI paused RL training on its latest deployment models for two weeks to harden and red-team infrastructure (@OpenAI)
X post · Story page
OpenAI says it paused RL training on its latest deployment-bound models for two weeks to harden and red-team its research infrastructure, citing stronger workload and network isolation, continuous security testing, and expanded multistage monitoring for higher-risk training.
Cursor is now a part of SpaceX (Cursor)
blog post · Story page
Cursor says it has been acquired by SpaceX and will use the combined compute to build more capable, lower-cost models for its coding agents, presenting Grok 4.6 as an early result.
GitHub Models is now retired (Simon Willison)
blog post · Story page
GitHub retired GitHub Models, the playground and unified model API that let Actions workflows run prompts on the GitHub API key already in the environment. Willison migrated his repository-summary workflow to OpenAI and bets the shutdown fits a pattern: coding-agent usage made free or subsidized tokens prohibitively expensive to offer.
swyx ships LLM-as-judge evals for the Kill My SaaS competition (@swyx)
X post · Story page
first LLM-as-judge evals for Kill My SaaS
Issues that covered it
- Daily
Maersk's 100,000 corrections, preference-trap evals, and Tencent's Hy4 Preview
Plus: guardrails at Navan, a one-way fiber, and the benchmark card that names its scaffold. 5 min.
Aug 31, 2026 · 5 min

- Daily
Aborted agent branches live on in the KV cache, plus StateM's harness scaling
Plus: coherence debt, the recall trap, Cline's open-weight evals. 5 min.
Aug 19, 2026 · 5 min

- Weekly #1
The score and the job came apart
Plus: a 1983 paper that explains this week, and a privacy leak nobody has confirmed. 8 min.
Aug 10 to Aug 16, 2026 · 8 min

- Daily
A public harness reproduces DeepSeek's 82.7% on Terminal-Bench, 445 trials deep
Plus: 4 talks from AI Engineer, a GitHub retirement, and sycophancy for smart people. 5 min.
Aug 10, 2026 · 5 min

Subscribe
Get the brief in your inbox
Pick daily, weekly, or both. Nothing is gated either way: every issue is on the site and in the feeds.
- Weekdays at 8:45am IST, one lead story and 6 to 9 items.
- Sundays, an argued synthesis rather than a recap.
- One click to leave, and quiet days say so in the subject line.



