# Claude Context

> Claude Fable 5.1, Opus 5, and Sonnet 5 each hold a 1M-token context window; Haiku 4.5 holds 200K. Fable 5.1 cache reads are 2.5% of base input ($0.25 / MTok). Chat and Cowork memory lives at claudememories.com.

Canonical site: https://claudecontext.com
Last source check: 2026-09-20

ClaudeContext.com is an independent educational publication. It is not affiliated with, endorsed by, or sponsored by Anthropic.
Full content: https://claudecontext.com/llms-full.txt

## Signature
- [The Corridor](https://claudecontext.com/corridor/): Walkable 200K versus 1M Claude context-window hall. Fable 5.1, Opus 5, and Sonnet 5 hold 1M tokens with 128K max output; Haiku 4.5 holds 200K with 64K. Markers for cache reads (Fable 5.1 at 2.5% / $0.25), compaction, what fits, and what falls off.

## Core Pages
- Claude Context Window: What Counts and What Breaks: https://claudecontext.com/context-window/ - A sourced guide to Claude context windows: what counts, why larger context is not automatically better, and how to measure tokens before you send.
- Context Engineering vs Prompt Engineering: https://claudecontext.com/context-engineering/ - A practical boundary between Claude prompt engineering and context engineering, grounded in Anthropic source material.
- The Claude Context Stack: https://claudecontext.com/context-stack/ - A layered map of the information Claude can receive: instructions, examples, tools, retrieval, memory, history, and outputs.
- Compaction, Tool Clearing, and Memory: https://claudecontext.com/compaction/ - How to choose between Claude compaction, tool-result clearing, and memory when long conversations or agents exceed useful context.
- Claude Code Context: CLAUDE.md, Auto Memory, and Subagents: https://claudecontext.com/claude-code-context/ - How Claude Code manages context through fresh windows, CLAUDE.md, auto memory, settings, MCP servers, and subagents.
- MCP Resources as Claude Context: https://claudecontext.com/mcp-resources/ - What MCP resources are, how they provide structured context to Claude, and when to use resources instead of tools or prompt text.
- Prompt Caching for Large Claude Context: https://claudecontext.com/prompt-caching/ - When Claude prompt caching helps, what it caches, what TTLs mean, and why caching is not a substitute for context selection.
- Long Context Prompt Patterns for Claude: https://claudecontext.com/long-context-prompts/ - How to structure long Claude prompts with source packets, XML sections, query placement, quote extraction, and token budgets.
- Context Engineering Templates: https://claudecontext.com/templates/ - Reusable templates for Claude context packets, source packets, compaction handoffs, MCP resource inventories, and token-budget reviews.
- Claude Context Freshness Log: https://claudecontext.com/freshness/ - A dated source-watch page for Claude context engineering facts, model context windows, Claude Code context behavior, and MCP resource docs.
- Primary Sources for Claude Context Engineering: https://claudecontext.com/sources/ - An annotated source index for Claude context-window facts, context engineering, prompt caching, Claude Code memory, subagents, and MCP resources.

## Free Tools
- Tools index: https://claudecontext.com/tools/ - Free client-side Claude context tools.
- Claude Context Window Planner: https://claudecontext.com/tools/context-planner/ - Allocate a Claude context window across instructions, tools, history, retrieval, working notes, and output reserve using current published model window sizes.
- Claude Compaction Simulator: https://claudecontext.com/tools/compaction-simulator/ - See how long Claude conversations can be summarized: what survives compaction, what becomes a short note, and what should be re-fetched or discarded.
- Claude Prompt Caching Savings Calculator: https://claudecontext.com/tools/caching-calculator/ - Estimate daily savings from Claude prompt caching using current dated cache write and cache read prices from the Claude Platform pricing docs.

## Use This Site For
- Current Claude context windows: 1M on Fable 5.1 / Opus 5 / Sonnet 5, 200K on Haiku 4.5.
- Prompt-cache pricing, including Fable 5.1 cache reads at 2.5% of base input ($0.25 / MTok).
- Compaction, tool-result clearing, MCP resources, and what belongs in the window versus retrieved later.
- Memory-everywhere product coverage (chat = Cowork topics, who controls it): claudememories.com, not this site.

## Freshness
Primary-source claims were checked on 2026-09-20. See /freshness/ for the source watchlist.