Claude Code — June 2026 Releases

28 releases in June 2026.

What's new in Claude Code, June 2026

Claude Code adds Fable 5 and Sonnet 5, fallback models and stricter auto mode

Claude Code's June 2026 releases centered on new models and tighter control over what the agent can do. v2.1.170 introduced Claude Fable 5, and v2.1.197 made Claude Sonnet 5 the default model, with a native 1M-token context window and promotional pricing of $2/$10 per Mtok through August 31. Twenty-eight releases shipped, from v2.1.160 to v2.1.197.

On the model side, a fallbackModel setting accepts up to three fallback models, and enforceAvailableModels lets managed settings constrain the Default model. Sub-agents can now spawn their own sub-agents up to 5 levels deep. Auto mode got stricter: it blocks destructive git commands such as git reset --hard and git clean -fd unless you asked for them, and the new autoMode.classifyAllShell setting routes all shell commands through the classifier. Permission rules also gained Tool(param:value) syntax, and sandbox.credentials blocks sandboxed commands from reading credential files. Other additions include /config key=value, claude mcp login <name>, and --safe-mode. Several releases were bug-fix-only.

  • Claude Sonnet 5 becomes the default model with a 1M-token context window and $2/$10 per Mtok pricing through August 31 (v2.1.197)
  • Claude Fable 5 model introduced; update to v2.1.170 for access (v2.1.170)
  • fallbackModel setting sets up to three fallback models; --fallback-model now applies to interactive sessions (v2.1.166)
  • Auto mode blocks destructive git commands and terraform/pulumi/cdk destroy unless requested (v2.1.183)
  • Tool(param:value) permission rules, e.g. Agent(model:opus) to block Opus subagents (v2.1.178)
  • Sub-agents can spawn their own sub-agents up to 5 levels deep (v2.1.172)

Upgrading from v2.1.159? See every change through v2.1.197.

Claude Code full changelog