Anthropic/OpenAI/Cursor/Gemini news, weekly roundups, ecosystem moves.
Anthropic acquired Stainless, the SDK auto-generation startup also used by OpenAI, Google, and Cloudflare. Reported price: $300M+. The next battleground is connection infrastructure.
Anthropic splits Claude Agent SDK billing into a separate pool on June 15. claude -p, GitHub Actions and SDK jobs get their own pool; interactive chat stays the same.
A single Claude Code changelog batch: Fast mode auto-upgrades to Opus 4.7, claude agents gains 8 CLI flags, worktree isolation can be turned off, and /plugin shows token cost before install.
Anthropic and PwC announced an expanded partnership on May 14. Claude Code and Cowork become standard tools across PwC’s global workforce — and a ready-made case for your own rollout.
Anthropic launched Claude Finance on May 5 — 10 agent templates covering IB, equity research, PE, wealth management, and analysis. The first vertical domain package.
Anthropic just bundled four slash commands that let Claude Code run on its own — /goal, /loop, /batch, /background. Here’s what changed and when to use each one.
Anthropic shipped four slash commands that change Claude Code’s default. Turn-by-turn pauses are no longer the only option — condition, interval, isolation, and parallel modes split the runtime.
After a year of daily Claude use, I’ve seen the same 5 prompting mistakes wreck answer quality. Fix these and output improves immediately.
Anthropic launched Claude for Legal with 20+ MCP connectors and 12 practice plugins. Google embedded Gemini Intelligence into Android, and GPT-5.5 Instant became ChatGPT’s default.
Claude Opus 4.7 launched with the same price tag but a tokenizer change pushing real costs 1.0–1.35x higher. Plus Claude Design, Cursor Canvases, Copilot 4.7 GA, and Qwen3.6 beating Opus on a local benchmark.