llms.txt
Compare with other sites' llms.txt files →# fable5.top
> Independent guide to Claude Fable5: specs, pricing math, access channels, prompt design guidance, templates, and dated blog analysis.
This site summarizes public information about Anthropic's Claude Fable5 model. It is not affiliated with Anthropic. Prefer official Anthropic documentation when site content conflicts with newer primary-source material.
## Primary
- [Claude Fable5 - Specs, Benchmarks & Pricing (July 2026)](https://fable5.top/): The independent guide to Claude Fable5: verified specs, honest benchmark analysis, $10/$50 pricing math, access channels, and prompting strategies.
- [Fable 5 Pricing - $10/$50 API & Max Plan Inclusion](https://fable5.top/pricing): Claude Fable5 API pricing: $10/$50 per million tokens, Max/Team Premium 50% weekly inclusion since July 20, 2026, usage credits on Pro, 90% prompt caching, and a calculator vs Opus 4.8 and Sonnet 5.
- [Claude Fable 5 Prompt Generator - Free System Prompt Builder & .xml Download](https://fable5.top/prompt-generator): Build an optimized Claude Fable 5 system prompt in seconds, then copy it or download the .xml. Set task type, autonomy, output style, and memory. Based on Anthropic's official guide.
- [Claude Fable5 Prompting Guide - Examples, Templates & Free Generator](https://fable5.top/claude-fable-5-prompt): How to write a Claude Fable5 prompt: what changed vs older Claude models, the XML anatomy that works, copy-paste examples, and a free interactive prompt generator.
- [Claude Fable5 Specs - Context Window, Effort & Model Card (2026)](https://fable5.top/claude-fable-5-specs): Every verified Claude Fable5 spec on one page: 1M-token context window, 128K max output, the effort parameter, removed sampling knobs, pricing, caching, and the Mythos 5 variant.
- [Claude Fable5 API - Access, Model ID & Code Examples (2026)](https://fable5.top/claude-fable-5-api): How to call the Claude Fable5 API: get a key, use model ID claude-fable-5, set the effort parameter, and choose between the Claude API, OpenRouter, and cloud platforms.
- [Claude Fable 5 Benchmarks (2026): SWE-Bench Pro 80.3%, Full Table & Honest Read](https://fable5.top/claude-fable-5-benchmarks): Every published Claude Fable 5 benchmark: 80.3% SWE-Bench Pro, FrontierCode, GDPval-AA, OSWorld - vs Opus 4.8, GPT-5.5 and Gemini 3.1, plus the footnote about which scores are really Mythos 5's.
- [Claude Sonnet 5 vs Fable 5 (July 2026): Price, Benchmarks & Which to Use](https://fable5.top/sonnet-5-vs-fable-5): Claude Sonnet 5 vs Fable 5 side by side: $2/$10 vs $10/$50 pricing, 1M context on both, benchmarks, availability after the July 1 restoration, and exactly when to pick each.
- [Fable 5 vs Opus 5 for Coding: Which Should You Use? (2026)](https://fable5.top/fable-5-or-opus-5-for-coding): A developer's decision guide: default to Opus 5 at high effort for everyday coding, escalate to Fable 5 for long-horizon agentic work, security-critical review, and 1M-context analysis. Benchmarks, prices and cost math.
- [When to Use Claude Fable 5 vs Opus 5 (August 2026): Worth It?](https://fable5.top/when-to-use-claude-fable-5): Is Fable 5 worth it after Opus 5 became the default? A decision tree for value/cost, compliance and long-context work, plus the 4 cases where Fable 5 still pays. Facts dated as of August 2026.
- [Claude Fable 5 in Claude Code: Select claude-fable-5, Setup & Gotchas](https://fable5.top/integrations/claude-code): Claude Code fully supports Claude Fable 5 via /model and the model selector. How to pick claude-fable-5, set effort, verify routing, and the silent-fallback gotcha. Verified August 2026.
- [Fable 5 vs Chinese Models (Qwen 3.8 Max, Kimi K3, GLM 5.2, DeepSeek V4) - as of 2026](https://fable5.top/fable5-vs-chinese-models): Claude Fable 5 vs the Chinese open-source cluster - Qwen 3.8 Max, Kimi K3, GLM 5.2, DeepSeek V4 Pro and Flash. Price, benchmarks, decision rows and export-control reality, dated as of August 2026.
- [Claude Fable 5 vs GLM 5.2: Open-Weights vs Frontier Coding (2026)](https://fable5.top/claude-fable-5-vs-glm-5-2): Fable 5 hits 80.3% on SWE-bench Pro at $10/$50; GLM 5.2 is ~7x cheaper on input with MIT open weights and a real coding score. Compare specs and benchmarks, then choose by task.
- [Claude Fable 5 vs DeepSeek V4 Pro: Price, Coding & Verdict (2026)](https://fable5.top/claude-fable-5-vs-deepseek-v4-pro): Fable 5 vs DeepSeek V4 Pro, as of August 2026: the cheap open-weights flagship ($0.435/$0.87, MIT, self-hostable) vs Anthropic's closed flagship. SWE-bench Pro 80.3% vs 55.4%, retention, export-control risk, and a per-task verdict.
- [Claude Fable 5 vs GPT-5.5 (August 2026): 2.5x Context, Best-in-Class Coding](https://fable5.top/claude-fable-5-vs-gpt-5-5): Fable 5 leads GPT-5.5 on every published coding/agentic benchmark (SWE-bench Pro 80.3% vs 58.6%) with 2.5x the context (1M vs 400K) - at $10/$50 vs $5/$30. Full spec table, sourced benchmarks, and a per-task decision guide.
- [Why Was Claude Fable 5 Unavailable? Ban Timeline & Restoration](https://fable5.top/claude-fable-5-unavailable): Claude Fable 5 was banned June 12 - July 1, 2026 under a US export-control directive. Full suspension timeline, why it happened, and what changed in the July 1 restoration.
- [Is Claude Fable 5 Back? Yes - Live Status (2026)](https://fable5.top/is-fable-5-back): Yes - Claude Fable 5 returned globally on July 1, 2026 after its export-control suspension. Live availability, rechecked every minute against Anthropic's status feed.
- [Claude Fable 5 Refund: How to Get Your Money Back 2026](https://fable5.top/claude-fable-5-refund): Fable 5 left paid Claude plans on June 22, 2026. Step-by-step refund guide for Claude Max/Pro, who qualifies, the deadline timeline, and cheaper alternatives.
- [Claude Fable 5 Alternatives: Best Free & Paid Models 2026](https://fable5.top/claude-fable-5-alternatives): The best Claude Fable 5 alternatives by task: Opus 5 ($5/$25 drop-in), Sonnet 5 ($2/$10), GLM 5.2 (free/open-source), GPT-5.5 and Gemini 3.1 - with migration steps.
- [Claude Fable 5 on AWS Bedrock - Access & Approval](https://fable5.top/claude-fable-5-aws): Access Claude Fable 5 on AWS Bedrock: identity verification, Anthropic's use-case approval, a copy-paste request template, and the setup steps to call it.
- [Is Fable 5 Included in Claude Max? Plans After July 20](https://fable5.top/claude-fable-5-max-plan): Claude Fable 5 on Max, Pro, and Team after July 20, 2026: permanent 50% weekly inclusion on Max/Team Premium, usage credits on Pro, and how that differs from API billing.
- [Fable5 Prompt Templates - 53 Copy-and-Customize Scaffolds](https://fable5.top/templates): 53 ready-to-paste prompts for Claude Fable5 - creative writing, research, coding, marketing, and more, plus 11 system prompt presets. Copy free.
- [Claude Code Tutorial: Install, Setup & First App (2026)](https://fable5.top/claude-code-tutorial): Step-by-step Claude Code tutorial with screenshots: install the CLI, use plan mode, generate CLAUDE.md, ship your first app. Timestamped to a beginner video.
- [Explore, Plan, Code, Commit: The Claude Code Workflow (2026)](https://fable5.top/claude-code-workflow): Anthropic's official Claude Code workflow with screenshots: plan mode, reviewing the plan, accept-edits, and committing. Timestamped to the source video.
- [Claude Code Subagents: Build a UI Review Agent (2026)](https://fable5.top/claude-code-subagents): What Claude Code subagents are and how to create one that reviews your UI in a browser: /agents, tools, and the review loop. Screenshots for every step.
- [Claude Code MCP Servers: Add Context7 & Playwright (2026)](https://fable5.top/claude-code-mcp): Add MCP servers to Claude Code: claude mcp add, --scope project, .mcp.json, the /mcp check, plus Context7 and Playwright walkthroughs - with screenshots.
- [Claude Code Hooks Guide: Events, Matchers & Blocking (2026)](https://fable5.top/claude-code-hooks): Learn Claude Code hooks through annotated screenshots: settings.json events and matchers, auto-formatting every edit, and blocking rm -rf or force pushes.
- [Claude Code Permissions & Allowed Tools Explained (2026)](https://fable5.top/claude-code-permissions): A screenshot walkthrough of Claude Code permissions: which built-in tools ask before running, the three-option edit prompt, how 'don't ask again' writes Bash(git add:*) into .claude/settings.local.json, and accept edits mode.
- [Claude Code Memory: CLAUDE.md and /init Walkthrough (2026)](https://fable5.top/claude-code-memory): How Claude Code memory works, with annotated screenshots: run /init to generate CLAUDE.md, edit it, save a rule with #, and open any memory file with /memory.
- [Claude Code Checkpoints and /rewind, Step by Step (2026)](https://fable5.top/claude-code-checkpoints): How Claude Code checkpoints and the /rewind command work, with annotated screenshots: Esc Esc to list them, the three restore scopes, and what a rewind will not undo.
- [Claude Code Custom Slash Commands: Step-by-Step (2026)](https://fable5.top/claude-code-slash-commands): Create custom slash commands in Claude Code: a markdown file in .claude/commands, front matter and $ARGUMENTS - your best prompt as one command. 14 screenshots.
- [GPT-6 Astra vs Claude Fable 5.1: Price & Benchmarks (2026)](https://fable5.top/gpt-6-astra-vs-claude-fable-5-1): GPT-6 Astra vs Claude Fable 5.1: same $10/$50 pricing, 1.05M vs 1M context, computer use and vendor benchmarks, September 2026. Which should you pick?
- [DeepSeek V4.1 Flash: Specs, Price & Self-Host (2026)](https://fable5.top/deepseek-v4-1-flash): DeepSeek V4.1 Flash: $0.15/$0.60 per 1M tokens off-peak, 2x during peak hours, 1M context and about 510 GB of MIT open weights. Spec sheet verified September 2026.
- [Claude Fable 5.1 vs DeepSeek V4.1 Flash: Price (2026)](https://fable5.top/deepseek-v4-1-flash-vs-claude-fable-5-1): DeepSeek V4.1 Flash vs Claude Fable 5.1: $0.15/$0.60 off-peak against $10/$50, 1M context both, MIT weights, cache $0.003 vs $0.25, Sept 2026.
- [Claude Fable 5.1 Prompt Caching: $0.25 Cache Reads (2026)](https://fable5.top/claude-fable-5-1-prompt-caching): Claude Fable 5.1 cache reads: $0.25/MTok, 0.025x input - a 75% cut. Write rates, TTL rules, DeepSeek $0.003 contrast and a worked 20-call example, Sept 2026.
- [Sakana Fugu Max Pricing & Specs: $2/$6 per 1M (2026)](https://fable5.top/sakana-fugu-max): Fugu Max costs $2/$6 per 1M tokens with $0.25 cache hits, fixed at any context length. Plus Fugu Ultra v2 pricing and what Sakana hasn't published.
- [Claude Skills Tutorial: Create, Run and Share Them (2026)](https://fable5.top/claude-code-skills): Create a Claude skill with the built-in skill creator, trigger it with /, install it in Claude Code via .claude/skills, and share it with your team. Screenshots for every step.
- [Claude Code Git Worktrees, Step by Step (2026)](https://fable5.top/claude-code-worktrees): Run parallel Claude Code sessions with git worktrees: claude --worktree, the .claude/worktrees layout, isolated subagents, and merging back to main - step by step.
- [Claude Code Context Management: /compact & /clear (2026)](https://fable5.top/claude-code-context): Read the /context meter, clear with /clear, compress with /compact, and understand auto-compact - what a compaction keeps, what it drops, and when to reset. With screenshots.
- [Claude Code GitHub Actions: @claude + PR Review (2026)](https://fable5.top/claude-code-github-actions): How to set up Claude Code in GitHub Actions: /install-github-app, the two workflows it installs, allowed_tools, @claude on an issue, and automated PR review.
- [Claude Code Output Styles: Built-ins and Custom Files (2026)](https://fable5.top/claude-code-output-styles): Output styles in Claude Code: the five built-ins with their picker wording, switching one in /config, and a custom .claude/output-styles file.
- [Claude Code Sandboxing: How to Enable the Bash Sandbox](https://fable5.top/claude-code-sandbox): Enable Claude Code's bash sandbox with screenshots: the bubblewrap install, the sandbox object in settings.json, the denyRead rule, and the network approval prompt.
- [Claude Code in VS Code: Setup Guide with Screenshots (2026)](https://fable5.top/claude-code-vscode): How to use Claude Code in VS Code: install the official Anthropic extension, open the side panel, and build and run an Express CRUD app - every step with screenshots from a real Claude Code 2.1.86 session.
- [Claude Code Plugins: Install, Manage and Build Your Own (2026)](https://fable5.top/claude-code-plugins): Claude Code plugins explained with screenshots: browse the official marketplace, install with claude plugin install, manage everything in /plugin, build your own plugin.json, and why the Anthropic verified badge matters.
- [Install Claude Code on Windows with WSL: Illustrated Guide (2026)](https://fable5.top/claude-code-windows): How to install Claude Code on Windows via WSL: enable WSL 2, run the curl install script inside Ubuntu, sign in with a Claude plan, trust your project folder, and check your usage with /cost - every step with screenshots.
- [Claude Opus 5.5 Pricing: $4/$20 per MTok, 1M Context (2026)](https://fable5.top/claude-opus-5-5): Claude Opus 5.5 costs $4 in / $20 out per MTok with a 1M-token context window by default and always-on adaptive thinking. Full specs, Fast mode rates, caching, Batch and how it compares with Opus 5 and Fable 5.1.
- [Claude Sonnet 5.5 Pricing: $2/$10 per MTok, 1M Context (2026)](https://fable5.top/claude-sonnet-5-5): Claude Sonnet 5.5 costs $2/$10 per MTok with a 1M context window, 128K output and a June 2026 cutoff - same price as Sonnet 5. Full pricing with cache and Batch, five breaking changes vs Sonnet 5, and the Opus 5.5 comparison.
- [Claude Sonnet 4.5 Deprecation: Retires November 30, 2026](https://fable5.top/claude-sonnet-4-5-deprecation): Anthropic deprecated Claude Sonnet 4.5 on September 30, 2026. It retires November 30, 2026 - see the model IDs, migration steps and what Sonnet 5.5 costs.
- [Claude Code Plan Mode: Plan, Approve and Ship Tasks](https://fable5.top/claude-code-plan-mode): Step-by-step Claude Code Plan Mode walkthrough with 16 screenshots: turn on plan mode, read a real implementation plan, pick the right approval option, handle Bash permissions, and use think / think hard / ultrathink.
- [Claude Code Statusline: /statusline Setup Step by Step](https://fable5.top/claude-code-statusline): How to set up the Claude Code statusline: run /statusline, pick the fresh default, approve the generated statusline-command.sh script and read the configured bar - every step illustrated from a real Windows session.
- [Claude Code JetBrains Plugin: IntelliJ IDEA Setup Guide (2026)](https://fable5.top/claude-code-jetbrains): Step-by-step screenshot guide to installing the Claude Code plugin in IntelliJ IDEA from the JetBrains Marketplace, fixing the 'claude is not recognized' error with the native installer, adding it to Path, signing in and running your first prompt.
- [Haiku 5.5 Benchmarks & Pricing: $0.10/$0.50, 1M Context](https://fable5.top/claude-haiku-5-5): Claude Haiku 5.5 benchmarks and pricing: $0.10/$0.50 per MTok, a 5x rate above 100K prompt tokens, 1M context and adjustable effort - plus Haiku 4.5 deltas.
- [Haiku 5.5 vs GPT-6 Luna (Luna 6): Price & Benchmarks](https://fable5.top/claude-haiku-5-5-vs-gpt-6-luna): Haiku 5.5 vs GPT-6 Luna: same $0.10/$0.50 base price, but above 100K prompt tokens Haiku jumps 5x while Luna 6 stays flat to 272K. Specs, benchmarks, cost math.
- [Claude Code Usage Limits and Cost Control: The Advisor Split](https://fable5.top/claude-code-usage-limits): Illustrated Claude Code cost-control walkthrough with 13 screenshots: attach an Opus advisor to a Sonnet executor with /advisor, compare it with /model opusplan, read the video's SWE-bench cost chart and see the same split in the Message API.
- [Claude Code on the Web: Browser Setup & Cloud Sessions](https://fable5.top/claude-code-web): Claude Code on the web: connect GitHub, set up a cloud environment, run /init, open the pre-filled pull request, then teleport the session to your terminal.
- [Claude Code Chrome DevTools MCP: Give It a Real Browser](https://fable5.top/claude-code-chrome): Give Claude Code a real Chrome with the Chrome DevTools MCP server: the README install line, the 26 tools it adds, and a debugging loop that names two failures.
- [Claude Fast Mode Pricing: Opus 5.5, 5 & 4.8 Rates (2026)](https://fable5.top/claude-fast-mode-pricing): Claude Fast mode bills Opus 5.5 at $8 in / $40 out and Opus 5 / 4.8 at $10/$50 per MTok for up to 2.5x more output tokens per second. Gated preview, rates, stacking rules.
- [Claude Opus 5.5 Prompt Hub - Copy-Paste Prompts & Video Workflows](https://fable5.top/prompts): Curated Claude Opus 5.5 prompts for coding, system prompts, and video or motion code. Copy the full prompt, see tools and workflow notes, and follow the source link. Free to use on fable5.top.
- [Claude Opus 5.5 Prompts - General Coding & System Prompts](https://fable5.top/prompts/claude-opus-5-5): Opus 5.5 prompts for coding, agents, and system prompts - copy-paste ready with workflow notes. Built for adaptive thinking and long-context work on Claude Max and the API.
- [Claude Opus 5.5 Video & Motion Prompts - Code-Driven Animation](https://fable5.top/prompts/claude-opus-5-5-video): Opus 5.5 prompts that generate HTML5 Canvas, WebGL, Remotion, or motion-code pipelines - with rendering tools, workflow notes, and copy-paste prompt text.
## Resources
- [Claude Fable5 Blog - Guides & Analysis (July 2026)](https://fable5.top/blog): Independent deep dives on Claude Fable5: pricing math, cost optimization, safeguard fallback mechanics, and practical prompting guides. Updated July 2026.
- [Kimi K3 vs Claude Fable 5.1: Open Weights vs the Updated Closed Flagship](https://fable5.top/blog/fable5-1-vs-kimi-k3): Moonshot Kimi K3 ($3/$15, open weights, strong Chinese) vs Claude Fable 5.1 (Sep 2 2026, SWE-bench 87.7%, $10/$50) - coding benchmarks, price, self-hosting, and when the Chinese-language flagship is the better choice.
- [Fable 5.1 vs DeepSeek for Coding: Benchmarks & Cost per Solved Task (2026)](https://fable5.top/blog/fable5-1-vs-deepseek-v4-pro-coding): Claude Fable 5.1 (Sep 2 2026, SWE-bench 87.7%, $10/$50) vs DeepSeek V4 Pro and V4 Flash for programming - agentic coding quality, open-weights self-hosting, and the cost-per-solved-task arithmetic updated for 5.1.
- [Claude Fable 5.1 国内怎么用?直连 API、中转接入与 2026 最新合规现状](https://fable5.top/blog/fable5-1-guonei-nengyong-ma): Claude Fable 5.1 正式发布后,国内开发者与企业该如何合法接入?本文详细梳理 Anthropic 官方直连条件、OpenRouter 等中转接入方式、30 天数据留存与出口管制合规现状,以及 Qwen 3.8 / GLM 5.3 国产平替选型。
- [Claude Fable 5.1 と GLM 5.3 を徹底比較:性能・2Mコンテキスト・料金・どちらを選ぶべきか(2026年9月)](https://fable5.top/blog/fable5-1-vs-glm-5-3): Anthropicの最新モデル Claude Fable 5.1 と Zhipuの GLM 5.3 を最新データで徹底比較。SWE-bench Verified、2Mコンテキストウィンドウ、API料金($10/$50 対 $1.20/$3.60)、オープンウェイト運用の違いを分かりやすく解説。
- [Claude Fable 5.1 Released: SWE-bench 87.7%, Adaptive Thinking & Full Benchmarks](https://fable5.top/blog/claude-fable-5-1-release-benchmarks): Anthropic released Claude Fable 5.1 on September 2, 2026. Here is the full breakdown of benchmarks (SWE-bench, FrontierMath, HLE), adaptive thinking token budgets, latency drops, and how it compares to Qwen 3.8 Max, GLM 5.3, and DeepSeek V4.
- [DeepSeek Harness (dsh): Plugins, Cordis Profiles, and the Community Directory (2026)](https://fable5.top/blog/deepseek-harness-plugins-guide): What DeepSeek Harness is as of August 2026 - Agent = Model + Harness, Cordis profile patches, how to install with npx @deepseek-ai/dsh web, and where to browse 488 community plugins synced from awesome-dsh-plugins.
- [Claude Sonnet 5 Pricing: the $2/$10 Intro Rate Is Now Standard](https://fable5.top/blog/sonnet-5-intro-pricing): Anthropic made Sonnet 5's $2/$10 introductory rate its standard price and cancelled the rise to $3/$15 scheduled for September 1, 2026. What that settles about routing and cost planni