Claude Code's 1M Context Window is Now Free: How to Use It Today

What Changed — The specific update

As of the latest Claude Code update, the 1 million token context window is now generally available and enabled by default for Claude Opus 4.6 and Claude Sonnet 4.6. This isn't a beta or a limited preview—it's the new standard.

The key detail for developers: Standard pricing applies across the entire 1M window with no long-context premium. This is a significant departure from competitors like OpenAI and Gemini, which charge more for prompts exceeding 200k-272k tokens.

According to Anthropic's announcement, Opus 4.6 scores 78.3% on MRCR v2 at 1 million tokens, making it the highest-performing frontier model at this scale. Media limits have also expanded to 600 images or PDF pages per request.

What It Means For Your Claude Code Workflow

This changes how you should approach large-scale coding tasks. Previously, you might have needed to chunk repositories or use complex retrieval strategies. Now you can:

1. Load entire codebases into a single context window
2. Process massive documentation sets without splitting
3. Run long-running agents that maintain full conversation history
4. Analyze complete project histories including multiple branches

No more worrying about hitting token limits mid-task or paying extra for long conversations. The 1M window is now your baseline.

Try It Now — No Configuration Needed

The update is automatic. When you use Claude Code with Opus 4.6 or Sonnet 4.6, you're already using the 1M context window. Here's how to verify and maximize it:

Check your version:

claude code --version

Load your entire project:

# Claude Code will now handle much larger codebases without truncation
claude code analyze --path ./your-entire-project

Create a CLAUDE.md that leverages the full context:

# Project Context

## Complete Architecture
[Include your entire architecture document here]

## All API Specifications
[Include all API docs]

## Full Codebase Structure
[Include directory tree and key file descriptions]

## Historical Context
[Include relevant git history, past decisions, team discussions]

Process large datasets:

# Analyze up to 600 PDF pages or images in one go
claude code process --files ./docs/*.pdf ./screenshots/*.png

What You Should Stop Doing

• Stop chunking large codebases unless absolutely necessary for organizational purposes
• Stop worrying about conversation length for complex refactors or debugging sessions
• Stop avoiding comprehensive documentation in your prompts—include it all

Performance Considerations

While the context window is larger, remember that extremely long contexts can still impact response times. For optimal performance:

1. Use /compact when you need faster responses on large contexts
2. Structure your prompts with clear sections and headings for better retrieval
3. Consider Sonnet 4.6 for cost-sensitive large-context tasks—it's significantly cheaper than Opus

This update fundamentally changes what's possible with Claude Code. You're no longer constrained by artificial token limits—your only constraints are now the actual capabilities of the models and your imagination.