Five times the memory, same price. Anthropic has rolled out 1M token context windows as a default feature on Claude MAX plans, removing what was previously an API-level upsell.
For context, Claude's standard context window sits at 200K tokens (roughly a 500-page book). The 1M token window expands that to approximately 2,500 pages of text in a single conversation. That means entire codebases, full legal contracts, or months of meeting transcripts can sit inside one session without the model "forgetting" earlier material.
What This Actually Changes in Practice
The practical difference between 200K and 1M tokens is bigger than the 5x multiplier suggests. At 200K, power users hit the wall constantly. You load a project, add some reference docs, have a back-and-forth conversation, and suddenly the model starts losing track of files you uploaded twenty minutes ago. The workaround has been breaking work into smaller chunks and re-feeding context, which burns time and subscription quota.
At 1M tokens, multi-hour work sessions become viable without context loss. Developers can keep entire repositories loaded. Researchers can feed in multiple papers and cross-reference them. Writers can maintain a full manuscript plus style guide plus revision notes in one thread.
How This Stacks Up
Google's Gemini 1.5 Pro has offered 1M token context since early 2025, and Gemini 2.5 Pro pushed that to 1M as well. OpenAI's GPT-4o tops out at 128K tokens. So this move puts Claude's consumer product on par with Gemini for raw context capacity, though the two models handle long-context retrieval differently in practice.
The key detail here is pricing structure. Previously, accessing Claude's extended context often meant paying per-token through the API. Bundling it into the flat-rate MAX subscription ($100/month) makes it predictable for heavy users who were watching API bills climb.
For anyone doing serious daily work in Claude, whether that is coding, research, or content production, this is the kind of upgrade that changes how you structure your workflow. Instead of planning around context limits, you can just work. That sounds small, but friction compounds. Removing it compounds too.