Claude Models Explained (2026): Fable, Opus, Sonnet, Haiku

Claude models come in a small family, and choosing between them is mostly about one trade-off: quality versus speed and cost. As of November 10, 2026, Anthropic's current lineup is Claude Fable 5.2, Claude Opus 5.5, Claude Sonnet 5.5 and Claude Haiku 5.5. Fable is the top tier for the hardest reasoning, Opus is Anthropic's recommended starting point for most workloads, Sonnet balances speed and intelligence, and Haiku is the fastest and cheapest.

This guide explains what each Claude model is for, how they differ, and how model choice works inside a Claude plan versus the API. It uses Anthropic's own descriptions from the official models overview and pricing docs, checked on November 10, 2026. It does not rank models with benchmark scores — for that, read Anthropic's own model pages, and test on your own tasks.

What "Claude models" means (and how it differs from Pro and Max)

Two things share the Claude name, and people mix them up:

• Models are the engines: Fable, Opus, Sonnet and Haiku. Each has a version number (5.2, 5.5) and a model ID used in code, such as claude-sonnet-5-5.
• Plans are subscriptions to Claude's apps: Free, Pro, Max, Ultra, Team and Enterprise. A plan decides which models you can pick in the app and how much you can use them.

So "Claude Pro" is not a model. It is a $20-a-month plan that gives you access to several models. Plan prices are in Claude Pro vs Max. Token prices for developers are in Claude API pricing.

The current Claude lineup at a glance

Anthropic's guidance on the same page: "If you're unsure which model to use, start with Claude Opus 5.5 for most workloads." Use Fable 5.2 for demanding reasoning and long-horizon agentic work, or when Opus 5.5 at higher effort still falls short in your own evaluations.

All four current models support text and image input, text output, multilingual use, vision and tool use, according to Anthropic.

Claude Mythos

Anthropic also lists Claude Mythos 5.2 on its pricing page, at the same API rates as Fable 5.2. It has limited availability: only organizations verified through Anthropic's verification programs, such as the Cyber Verification Program, can use it. Most people will never select it.

Fable: the top tier for hard reasoning

Claude Fable 5.2 is Anthropic's model "for demanding reasoning and long-horizon agentic work". It is the most expensive current model on the API at $8 input / $40 output per million tokens, and its thinking is always on.

When Fable makes sense:

• Multi-step agent runs that go on for a long time and must stay on track.
• Hard analytical or technical problems where Opus is not getting there in your tests.

When it does not:

• Everyday writing, chat and summaries — you pay more and wait longer for little gain.

On plans: Fable is not on Free. Pro users pay for it with usage credits. Max and Ultra include Fable up to half the weekly limit, per Anthropic's pricing page.

Opus: the recommended default for most workloads

Claude Opus 5.5 is described as "for long-running agentic coding and knowledge work", and it is where Anthropic tells API developers to start if unsure. On the API it costs $4 input / $20 output per million tokens, with cache reads at just $0.20.

Opus is a good fit for:

• Coding agents and multi-file changes (it is a common choice in Claude Code).
• Long documents, research synthesis and complex knowledge work.
• Agentic workflows that call tools many times.

Its default effort setting is medium on the API, and thinking is always on. Fast mode is available for Opus 5.5 at a premium ($8 / $40) when speed matters.

On plans: Opus is included on Pro and higher; Free users do not get Opus.

Sonnet: speed plus intelligence

Claude Sonnet 5.5 is "the best combination of speed and intelligence". It is faster than Opus and costs half as much on the API: $2 input / $10 output, with cache reads at $0.10.

Sonnet is a strong choice for:

• Production chat features and customer-facing assistants where latency matters.
• Drafting, editing and summarizing at volume.
• Tool use where you need a quick answer, not a long deliberation.

Its default API effort is high. On plans: Sonnet is available on every plan, including Free.

Haiku: speed and cost

Claude Haiku 5.5 is for "high-volume, latency-sensitive tasks such as classification, extraction, and routing". It is the fastest current model and the cheapest: $0.10 input / $0.50 output per million tokens for prompts up to 100,000 tokens, rising to $0.50 / $2.50 for prompts above that.

Use Haiku for:

• Labeling, tagging and routing large numbers of items.
• Pulling structured fields out of text.
• A first, cheap pass before sending hard cases to Sonnet or Opus.

On plans: Haiku is available on every plan, including Free.

How model choice works in a Claude plan vs the API

In Claude's apps (a plan): you pick a model from the model picker. Heavier models use up your plan's limit faster. Anthropic's pricing page says how fast you reach a limit "depends on the model you use, how long your chats are, and how large your files are". If you keep hitting limits on Pro, switching everyday tasks to Sonnet or Haiku stretches your usage.

On the API: you choose a model per request with its model ID and pay per token. You can mix models in one app — for example, Haiku to route, Sonnet to answer, Opus for the hard cases. Full rates, caching and batch discounts are in Claude API pricing.

One tokenizer note: Anthropic says Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text than earlier models. Count tokens on the model you will actually use.

Older Claude models

Anthropic keeps several legacy models available: Fable 5.1, Fable 5, Opus 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 4.5, Sonnet 5, Sonnet 4.6 and Haiku 4.5. Each has a retirement commitment on Anthropic's deprecations page; for the current four models, Anthropic commits to retirement no sooner than September–November 2027 depending on the model. If you are on an older model, check the price table — several newer models are cheaper per token than the ones they replace.

Related tools: Claude Code

Claude Code, Anthropic's coding agent, runs on these same models and is included on paid plans. It is not a separate model. For which plan suits coding work, see Claude Code pricing. To compare Claude with other assistants, see Claude vs ChatGPT, Claude vs Gemini and our ChatGPT alternatives guide.

FAQ