Claude Code Gets 1M Context Window, Cowork Projects, and Channels in Latest Updates

Claude AI code editor interface with project folders and warm orange-brown accent

Anthropic has shipped a series of significant updates to Claude Code, its CLI tool for developers, over the past week. The highlights include a massive context window expansion, the launch of Projects in the Cowork desktop app, and experimental features that hint at where AI-assisted development is heading.

1M Context Window for Opus 4.6

The biggest technical upgrade: Claude Opus 4.6 now gets a 1 million token context window by default for Max, Team, and Enterprise plan users. Previously, this required extra usage allocation. This means Claude Code can now process entire codebases, lengthy documentation, and complex multi-file projects without hitting context limits.

Additionally, default maximum output token limits for Opus 4.6 have been increased to 64k tokens, with an upper bound of 128k for both Opus 4.6 and Sonnet 4.6 models.

Projects in Cowork

The Cowork desktop app now supports Projects — letting developers keep tasks and context in one place, focused on a single area of work. Files and instructions stay on your computer. You can import existing projects in one click or start fresh. This makes the Cowork experience much more organized for developers working across multiple codebases.

Channels (Research Preview)

A new experimental feature called Channels allows MCP servers to push messages directly into your Claude Code session. This opens the door for real-time integrations — imagine getting CI/CD notifications, deployment status updates, or monitoring alerts directly in your coding session without polling.

Other Notable Updates

  • Rate limit visibility — New rate_limits field in statusline scripts shows Claude.ai rate limit usage (5-hour and 7-day windows with percentage and reset times)
  • /color command — Set a prompt-bar color for your session to visually distinguish between projects
  • /copy N — Copy the Nth-latest assistant response instead of just the last one
  • --bare flag — For scripted -p calls, outputs just the result without formatting
  • allowRead sandbox setting — Re-allow read access within denyRead regions for more granular file permissions
  • Bug fix — "Always Allow" on compound bash commands now saves per-subcommand rules instead of one rule for the entire string

The Bottom Line

The 1M context window is the headline feature — it effectively removes the biggest limitation of AI-assisted coding for large projects. Combined with Projects in Cowork and the Channels experiment, Anthropic is clearly building toward a future where Claude Code isn’t just a coding assistant but a persistent development companion that stays connected to your entire workflow. The pace of these updates is impressive.