Claude Fable vs Opus is the choice most Claude users face now, and Anthropic's own advice is simple: start with Claude Opus 5.5 for most work, and move up to Claude Fable 5.2 only for the most demanding reasoning and long running agentic tasks. Claude Mythos 5.2 is not a third option for most people. It is the same model as Fable 5.2 with looser safeguards, available only to vetted cybersecurity and life science organizations. On Mythos vs Opus 5.5: Opus 5.5 is the model you can actually buy at half Fable's token price, and Anthropic still calls it the best place to start.

This guide covers the whole lineup, plus Sonnet and Haiku 5.5, with prices, plan access and fast mode details rechecked on November 10, 2026.

What's new (November 10, 2026): Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9. It costs $8 and $40 per million tokens (Fable 5.1 was $10 and $50), runs up to 30% faster and gets fast mode. Fable 5.1 is now a legacy model (release notes).

The current Claude lineup

Anthropic now sells four model lines through the Anthropic API and the Claude apps: Fable at the top, then Opus, Sonnet and Haiku. The 5.5 generation started in September, and Claude Haiku 5.5 joined it on October 7.

ModelReleasedAPI price per million tokens (input / output)Cache readsSpeedContext
Claude Fable 5.2Nov 9, 2026$8 / $40$0.20Slower1M tokens
Claude Opus 5.5Sep 22, 2026$4 / $20$0.20Moderate1M tokens
Claude Sonnet 5.5Sep 28, 2026$2 / $10$0.10Fast1M tokens
Claude Haiku 5.5Oct 7, 2026From $0.10 / $0.50From $0.01Fastest1M tokens

In plain terms: Fable 5.2 costs eight dollars per million input tokens and forty per million output tokens, down from ten and fifty for Fable 5.1. Opus 5.5 costs four and twenty, Sonnet 5.5 two and ten, and Haiku 5.5 ten cents and fifty cents for prompts up to 100,000 tokens (fifty cents and two fifty above that). All four current models handle up to a million tokens of context and up to 128,000 output tokens. These figures come from Anthropic's models overview and pricing page.

Claude Fable vs Opus: what is the real difference?

Fable 5.2 is Anthropic's most capable generally available model. It always uses adaptive thinking and is aimed at research, large engineering jobs and work that runs for hours unattended. In Anthropic's Fable 5.2 announcement, it scores 49.1% on FrontierCode 2.0 at xhigh effort against 45.5% for Opus 5.5, and at high effort it roughly matches Opus 5.5's best score for about half the cost per task. On the independent Artificial Analysis leaderboard, Fable 5.2 (max) ranked first at 65 when we checked on November 10, ahead of Gemini 4 Carbon (high) at 60 and Opus 5.5 (max) at 59.

Opus 5.5 is the newer, cheaper workhorse. Anthropic says it "performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5," and it generates output more than 30 percent faster than Opus 5, per the Opus 5.5 launch post.

Anthropic recommends Fable when your own tests on Opus 5.5, even at higher effort, still fall short, such as long horizon research and the hardest debugging. Feels weaker lately? See is Claude Opus 5.5 nerfed.

Pros of Fable 5.2: strongest Claude for open ended, multihour work; 20% cheaper than Fable 5.1; fast mode now available. Cons of Fable 5.2: twice Opus 5.5's token price, slower, metered on consumer plans, and 30 day data retention by default.

Pros of Opus 5.5: near-Fable quality at much lower cost, faster, fast mode available, Anthropic's recommended default. Cons of Opus 5.5: a few breaking API changes for teams upgrading from Opus 5, such as thinking that can no longer be turned off.

Claude fast mode: what it is and what it costs

Claude fast mode is a faster way to run Opus or Fable, not a separate model. Anthropic's fast mode documentation says it runs the same model with a faster inference configuration, with "no change to intelligence or capabilities", and delivers up to 2.5 times more output tokens per second. The gain is in output speed, not in time to the first token, so it shows most when you stream long answers.

ModelFast mode input / output per million tokensStandard priceAccess
Claude Fable 5.2$16 / $80$8 / $40Available from launch, no beta header
Claude Opus 5.5$8 / $40$4 / $20Generally available, no beta header
Claude Opus 5 and Opus 4.8$10 / $50$5 / $25Research preview with beta header; Opus 4.8 fast mode is deprecated and ends Nov 30, 2026

Prices are from Anthropic's pricing page. On the API you turn it on by setting speed: "fast" on a request. A few limits matter:

  • It works on the Claude API, Claude Models on Google Cloud and Microsoft Foundry, but the docs do not list Amazon Bedrock.
  • It is not available with the Batch API, it has its own rate limit, and Sonnet and Haiku do not offer it. Mythos 5.2 fast mode is only for verification program members.

In Claude Code you toggle it with /fast. The Claude Code fast mode page says subscribers pay for it from usage credits, not plan limits, and turning it on midway charges the fast input price for the whole context once, so switch it on at the start.

Our take: use fast mode for live debugging and quick back and forth where waiting costs you more than tokens. Leave it off for batch jobs and long autonomous runs, where it doubles the bill for time you are not watching.

Where Sonnet 5.5 fits

Claude Sonnet 5.5, released September 28, is the speed and cost option. Anthropic says it is at least 30 percent faster than Sonnet 5 and costs up to 30 percent less per task. On some of Anthropic's own benchmarks it comes close to or even beats Opus 5.5, but the company is explicit that Opus 5.5 "remains clearly stronger at complex, open ended work requiring sustained judgment." Use Sonnet 5.5 for well scoped tasks, bug fixes, documents and slides, and high volume workloads.

Where Haiku 5.5 fits

Claude Haiku 5.5, released October 7, is Anthropic's fastest model at standard speed and its cheapest current model. It is aimed at high volume, narrowly scoped work such as classification, summaries, compaction and subagent tasks under Opus or Sonnet, and Anthropic says Sonnet 5.5 and Opus 5.5 remain better for complex agentic coding. It is also the first Haiku with an adjustable effort setting. For how it compares with OpenAI's equally priced small model, see Claude Haiku 5.5 vs GPT-6 Luna.

Claude Mythos vs Fable: same model, different guardrails

Mythos is the name for Anthropic's frontier model with fewer safeguards. With the 5.2 release, "Claude Fable and Claude Mythos are the same underlying model," in Anthropic's words, and Fable carries the stronger cyber and biology safeguards.

  • Fable routes several dual use cyber tasks, such as penetration testing, exploit generation and binary vulnerability scanning, to Opus models, and sends life science research and development questions to Opus as well. It can be used to find software vulnerabilities, but not to develop exploits. Anthropic says Fable 5.2's safeguards are as strong as 5.1's but block fewer harmless requests.
  • Mythos 5.2 relaxes those limits for vetted users through two programs: the Cyber Verification Program for defensive security work, and the Life Sciences Verification Program, built with the US government, for professional biology research. Anthropic says both programs were expanded for 5.2. Claude Enterprise customers can also reach Mythos 5.2 without a program through Claude Security, Anthropic's code scanner.

Mythos vs Opus 5.5: which one can you use?

Mythos 5.2 and Opus 5.5 are not really rivals. Mythos 5.2 is Fable 5.2 with relaxed safeguards for vetted defensive security and life science users, while Opus 5.5 is the general model Anthropic recommends as the default for everyone else.

  • Capability. Anthropic's benchmark table puts Fable 5.2 and Mythos 5.2 "a step above Opus 5.5 in almost all the capabilities we evaluated," with the gap largest on the hardest, longest tasks.
  • Access. Opus 5.5 is on Pro, Max, Team and Enterprise plans and the Anthropic API. Mythos 5.2 needs approval through the Cyber Verification Program or the Life Sciences Verification Program.
Mythos 5.2 vs Opus 5.5Claude Mythos 5.2Claude Opus 5.5
API price per million tokens$8 input, $40 output$4 input, $20 output
Who can use itVetted cyber and life science organizationsAnyone on a paid plan or the API
CVE Rediscovery (Anthropic test)89.6%81.3% (without safeguards)
Terminal-Bench-Science 0.175.3%58.7%
Sandbox escape test (lower is better)3.1% of scenarios3.3% of scenarios

Figures are Anthropic's own tests from its 5.2 launch post. In words: Mythos 5.2 costs twice as much, found 10.2% more known vulnerabilities and scored far higher on science tasks, while sandbox behavior was about the same. On AI research work such as kernel optimization, Anthropic held it to about Opus 5.5's level.

Outside those programs, pick Opus 5.5 or Fable 5.2. For a rival flagship, see our Muse Wonder explainer.

Safety testing behind Opus 5.5

Opus 5.5 was tested before release by outside evaluators including METR, and it ships with the same safeguards as Fable. For what Anthropic does with flagged conversations once a model is in use, see whether Claude reports you to the police; our Anthropic IPO explainer covers governance.

Which Claude model is on your plan?

According to Claude's pricing page:

  • Free: Sonnet and Haiku.
  • Pro ($20 a month, or $17 a month billed annually): adds Opus; Fable is available through usage credits.
  • Max ($100 a month, five times Pro usage) and Ultra ($200 a month, ten times Pro's weekly usage): Opus included; Fable can use up to half of your weekly limits.
  • Team premium seats and Enterprise: Fable available; premium seats get up to half of weekly limits.

For the 5.2 launch week, Anthropic says eligible paid plans get one to three Fable resets, and Pro plans and standard Team seats can use Fable for up to half of their weekly limits (launch post).

How to choose

  1. Default to Opus 5.5 for coding, analysis and writing.
  2. Use Sonnet 5.5 when speed or cost matters more than the last few points of quality.
  3. Use Haiku 5.5 for bulk, simple or real time tasks, and as a cheap subagent.
  4. Escalate to Fable 5.2 for long, unattended or research grade work where Opus 5.5 falls short in your tests.
  5. Apply for Mythos only if you are a vetted security team or life science organization hitting Fable's safeguards.

To see how Claude's top model stacks up against OpenAI's, read GPT-6 Astra vs Claude Fable 5.1. For the safety policy behind the Fable and Mythos split, see Preparedness Framework vs Responsible Scaling Policy, and for the plan prices, ChatGPT Plus vs Claude Pro vs Google AI Pro.

From Ahzy Capital (our own products, labeled as such): Get the free AI Agent Cheat Sheet: 24 copy and paste AI agent setups for 8 types of small business, with prompts and workflows. We make and sell these, so they aren't independent picks.

Spec sheets: Fable 5.1 (the previous Fable), Opus 5.5, Sonnet 5.5, Haiku 5.5 and the release timeline.

FAQ

Is Claude Fable better than Opus?

Fable 5.2 is the more capable model and led the Artificial Analysis index on November 10, but it costs twice as much per token as Opus 5.5. Anthropic recommends Opus 5.5 first and Fable for the hardest tasks.

Mythos vs Opus 5.5: which is better?

Mythos 5.2 is stronger: it is Fable 5.2 with relaxed safeguards, and on Anthropic's CVE Rediscovery test it found 10.2% more vulnerabilities than Opus 5.5. But only vetted organizations can use it, at $8 and $40 per million tokens. Opus 5.5 costs $4 and $20 and is open to anyone on a paid plan or the API, so for most people it is the better choice.

Is Claude Mythos available to the public?

No. Claude Mythos 5.2 is available only to vetted organizations through Anthropic's Cyber Verification and Life Sciences Verification programs. The public version of the same model is Claude Fable 5.2.

How much does Claude Fable cost?

Fable 5.2 costs $8 per million input tokens and $40 per million output tokens on the API, with cache reads at $0.20 per million, or $16 and $80 in fast mode. On Claude Pro it is accessed through usage credits, and on Max and Ultra it can use up to half of the weekly limits.

What is Claude fast mode?

It is a faster configuration of Claude Opus or Fable with the same model quality and up to 2.5 times the output speed, at double the standard price: $8 and $40 per million tokens on Opus 5.5, $16 and $80 on Fable 5.2. In Claude Code you switch it on with /fast.

Should I use Claude Opus 5.5 or Sonnet 5.5?

Use Opus 5.5 for complex, open ended work that needs sustained judgment. Use Sonnet 5.5 for faster, well scoped tasks and high volume use; it costs half as much per token.