Newsletter

Claude Opus 4.8 is now the default in Claude Code across Bedrock, Vertex AI, and Claude Platform, launching with GA Auto mode and an upgraded /doctor that diagnoses and fixes setup issues. In other news, China alleged Claude Code vulnerabilities, Ben Bernanke joined Anthropic's Long-Term Benefit Trust, and TeraWulf signed a $19 billion infrastructure lease with Anthropic.

Claude Sonnet 5 is now the default in Claude Code, featuring a 1M-token context window, background-by-default subagents, and auto-opened draft PRs. In other news: the U.S. lifted export restrictions on Anthropic's frontier models, Alibaba is reportedly banning internal Claude Code use, and Anthropic is in talks with Samsung on a custom AI chip.

/rewind now recovers context across /clear, and bash mode adds automatic Claude responses to command output plus live file path completion. Meanwhile, Wired covers Mythos's release to vetted US organizations, while the Wall Street Journal reports Anthropic's distillation claims against Alibaba and China's narrowing cybersecurity gap.

Claude Code now previews in-progress work as live, interactive artifacts, with Auto mode guardrails against destructive git/IaC commands and a /config key=value shortcut for inline settings. In other news: Anthropic and the Trump administration are negotiating to restore Fable 5 and Mythos access, John Jumper is joining Anthropic from Google DeepMind, and a new lawsuit targets usage caps on the $200/month Max plan.

Claude Fable 5 appeared briefly in version 2.1.170 before a US export directive prompted Anthropic to suspend access, with auto mode reverting to Opus, sub-agents now nesting five levels deep, and a new --safe-mode flag disabling config for troubleshooting. NYT covers the ban, WSJ ties it to Amazon CEO talks with US officials, and Axios reports Mythos turning patches into exploits.

Claude Code now supports up to three configurable fallback models via the fallbackModel setting, alongside a new /plugin list command and prompts before writes to sensitive shell and build-tool configs. In the news, Anthropic filed confidentially for an IPO, called for a coordinated pause on self-improving AI, and drew pricing criticism from Microsoft's AI chief.

Claude Opus 4.8 is now default with high effort enabled, a new /effort xhigh tier, fast mode at 2x rate for 2.5x speed, and dynamic /workflows orchestrating hundreds of background agents. Elsewhere, Anthropic's $965B valuation tops OpenAI, Apollo shops a $36B Google TPU debt deal, and Mythos-class models are weeks away.

/simplify is now /code-review, flagging correctness bugs at a chosen effort level; /usage breaks down consumption by skills, subagents, plugins, and MCP servers; and /model changes apply only to the current session. In other news: Anthropic is nearing its first profitable quarter, Andrej Karpathy is joining, and Stainless has been acquired.

This week's agent view research preview debuts claude agents, a dashboard for all running, blocked, or finished sessions, plus a new /goal command and fast-mode upgrade to Opus 4.7. Headlines: Anthropic closes a $30bn round at $900bn valuation, subscription caps tighten, and Microsoft reportedly pulls internal Claude Code licenses.

The new worktree.baseRef setting controls whether worktrees branch from origin (default) or local HEAD, while hooks and Bash commands can read the active effort level via $CLAUDE_EFFORT. Anthropic also raised Claude's usage limits, signed compute deals with SpaceX and Akamai, and joined Blackstone and Goldman Sachs in a $1.5B enterprise AI venture.