Claude models explained: Fable, Mythos, Opus, Sonnet and Haiku — which to use

Claude now comes in five model families, and the names are easy to mix up. Fable and Opus are the powerful, pricier models, Sonnet is the balanced everyday one, Haiku is the fast, cheap one, and Mythos is a restricted version of Fable for vetted professionals. This guide sets out what each model is for, what it costs in the API, which plans include it and how to choose, using Anthropic’s model docs and announcements as of September 2026.

Claude’s current models at a glance

Anthropic rates their speed as slower for Fable 5.1, moderate for Opus 5.5, fast for Sonnet 5.5 and fastest for Haiku 4.5. Claude Sonnet 5.5 launched on September 28, 2026, and Claude Haiku 5.5 is due in the coming weeks, with many of the same improvements as Opus 5.5.

Fable 5.1 and Mythos 5.1

Anthropic launched Fable 5.1 and Mythos 5.1 on September 1, 2026, calling them its most advanced models for coding and knowledge work. It says they are “the same model, but with different levels of safeguards”:

• Fable 5.1 is generally available with standard safeguards. It can identify software vulnerabilities but not develop exploits, and it handles elementary biology and medical questions itself while redirecting research and development queries to Opus models. Anthropic says it produces 85% fewer false positives on biology and medical questions and 60% fewer on cybersecurity than the earlier version.
• Mythos 5.1 has reduced safeguards for vetted professionals. It is limited to US organisations for now, through the Cyber Verification Program for defensive security professionals and the Life Sciences Verification Program for researchers, and it also powers Claude Security.

Fable 5.1 costs $10 and $50 per million tokens, and Anthropic says that is about 25% less than Fable 5 for typical workloads and up to 45% less for highly agentic tasks.

Opus 5.5

Claude Opus 5.5, launched on September 22, 2026, is the first model in the Claude 5.5 family. Anthropic says it performs at the level of Fable 5.1 on most work and costs 40% less to run than Opus 5. Its docs recommend it as the place to start for most workloads. Read our report on Claude Opus 5.5 for the pricing, benchmarks and plan-limit changes.

Sonnet 5.5 and Haiku 4.5

Claude Sonnet 5.5, launched on September 28, 2026, is a faster, lower-cost complement to Opus 5.5, with a 1M-token context window at $2 and $10 per million tokens, output more than 30% faster than Sonnet 5 and up to 30% less cost per task. Read our report on Claude Sonnet 5.5. It replaces Claude Sonnet 5, which launched on June 30, 2026 at the same price. Claude Haiku 4.5 is the fastest and cheapest, with a 200K-token context window. Anthropic says Haiku 4.5 won’t be retired sooner than October 15, 2026, so check its model page if you build on it.

Which plans include which models

Our guide to Claude Pro vs Max explains the limits and how Fable is billed on each plan.

Which Claude model should you use?

• Most work, writing, research and coding: Opus 5.5, per Anthropic’s own guidance.
• The hardest reasoning or very long agent runs: Fable 5.1, or try Opus 5.5 at a higher effort first to save money.
• Fast, high-volume or cost-sensitive tasks: Sonnet 5.5, or Haiku 4.5 for the quickest answers.
• Professional security or life sciences research: apply for Mythos 5.1 through the verification programs.

For coding with these models, see how to use Claude Code and the best AI coding assistants. To compare with OpenAI’s lineup, read ChatGPT models explained, and for the assistant overall, the Claude profile.