What's new in Claude

Session titles now generate in the conversation's language; improved Bedrock credential caching; fixed model enforcement for aliases and fast-mode toggling; resolved numerous bugs in auto mode, Remote Control, background sessions, and UI rendering across platforms.

Added enforceAvailableModels managed setting, giving admins the ability to constrain the Default model selection and prevent widening of managed model allowlists.

Added mouse-wheel scroll acceleration toggle; fixed model picker display issues for various plan types; improved Bedrock GovCloud support; fixed background session environment variable inheritance.

Claude Opus 4.8 is now the default model for Max, Team Premium, Enterprise pay-as-you-go, and API accounts, with high effort by default and /effort xhigh available for the hardest tasks.

Dynamic workflows: Claude now writes its own orchestration harness on the fly, coordinating dozens to hundreds of subagents from a script it composes custom-built for each task.

Security-guidance plugin: a new plugin from the marketplace (/plugins) reviews Claude's changes for vulnerabilities in real time as it writes code — available to all Claude Code users.

Fast mode on Opus 4.8 now runs at $10/$50 per MTok, giving Claude Code users faster Opus responses at fixed pricing for interactive and background sessions.

Fable 5 and Mythos 5 use the tokenizer introduced with Opus 4.7. The same text now produces roughly 30% more tokens than pre-Opus-4.7 models — use the token counting API with model: "claude-fable-5" to measure prompts under the new tokenizer.

The GET /v1/environments/{id}/work endpoint for self-hosted sandboxes is now available on Claude Platform on AWS, with the new GetEnvironment IAM action authorizing access.

Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage — covering the Agent SDK, claude -p, GitHub Actions, and third-party Agent SDK apps. Weekly limits also up 50% through July 13 for all Pro, Max, Team, and Enterprise users.

Claude Code now stretches across terminal, IDE, web, desktop, and scheduled agent workflows. Managed Agents can run in your own sandbox connected to private MCP servers. 1M token context window enabled on Opus 4.6 and Sonnet 4.6 in Claude Code — no beta header, no surcharge, flat pricing.

"I haven't written a line of code myself in 2026." Boris Cherny, Claude Code's creator, recently led the launch of the plugins system — the easiest way to share and install bundles of agents, slash commands, MCP servers, and hooks across teams.

Thariq shared a deep-dive on dynamic workflows: Claude can now write its own harness on the fly, custom-built for the task at hand — making multi-agent orchestration a first-class part of every session.