Anthropic's current Claude models are Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5 and Claude Haiku 5.5, and Anthropic's own advice is to start with Opus 5.5 for most workloads. As of Nov 2026 all four have a 1M token context window, 128K tokens of max output and a reliable knowledge cutoff of Jun 2026, so the real differences are speed, price and how hard each one thinks.
This is the lineup as Anthropic's developer docs lay it out, which is the API view. Each model's page in those docs lists the platforms it's available on. All four handle text and image input, text output, multiple languages, vision and tool use.
The four Claude models in plain words
- Claude Fable 5.1: for demanding reasoning and long-horizon agentic work. Anthropic rates it Slower.
- Claude Opus 5.5: for long-running agentic coding and knowledge work. Moderate speed, and the default pick.
- Claude Sonnet 5.5: the best combination of speed and intelligence. Fast.
- Claude Haiku 5.5: for high-volume, latency-sensitive tasks such as classification, extraction and routing. Fastest.
Anthropic's guidance on Fable is specific: use it for demanding reasoning and long-horizon agentic work, or when your tests on Opus 5.5 at higher effort still fall short. In other words, Fable is the step up when Opus can't do it, not the place to begin.
Side by side: speed, price and limits
Base prices are per million tokens and copied from Anthropic's docs as of Nov 2026.
| What the docs list | Fable 5.1 | Opus 5.5 | Sonnet 5.5 | Haiku 5.5 |
|---|---|---|---|---|
| Best for | Demanding reasoning, long agentic work | Long-running coding and knowledge work | Speed plus intelligence | High-volume sorting and extraction |
| Speed | Slower | Moderate | Fast | Fastest |
| Input price | $10 | $4 | $2 | From $0.10 |
| Output price | $50 | $20 | $10 | From $0.50 |
| Context window | 1M tokens | 1M tokens | 1M tokens | 1M tokens |
| Max output | 128K tokens | 128K tokens | 128K tokens | 128K tokens |
| Thinking | Adaptive, always on | Adaptive, always on | Adaptive | Adaptive |
| Default effort | High | Medium | High | Medium |
| Retirement | Not sooner than 1 September 2027 | Not sooner than 22 September 2027 | Not sooner than 28 September 2027 | Not sooner than 7 October 2027 |
Haiku's price has a catch the summary table hides. On the pricing page, Haiku 5.5 is $0.10 input and $0.50 output for prompts up to 100,000 tokens, and $0.50 input and $2.50 output for prompts over that. Each request is priced on its own, so one very long prompt pays the higher rate even when part of that prompt comes from a cache hit.
Two discounts apply across the lineup. Batch API requests are half price, and prompt cache reads cost a fraction of the base input price; on Opus 5.5 a cache hit is $0.20 per million tokens. Choosing US-only inference through the inference_geo setting adds a 1.1x multiplier to every token price.
What 1M tokens means in words
Anthropic says 1M tokens is roughly 555k words on the current tokenizer, introduced with Claude Opus 4.7. Models before it fit about 750k words in the same 1M tokens. So if you're comparing an old model with a new one, the same document uses more tokens on the newer models. Max output, the 128K figure, is the limit for a single synchronous reply.
Adaptive thinking and effort, minus the jargon
Adaptive thinking means the model decides how much to think before answering, steered by a setting called effort. On Fable 5.1 and Opus 5.5, thinking is always on. The default effort is high on Fable 5.1 and Sonnet 5.5, and medium on Opus 5.5 and Haiku 5.5, and a developer can set it differently on each request.
The older manual mode, called extended thinking, where you set a thinking budget by hand, is deprecated on Claude Opus 4.6 and Claude Sonnet 4.6 and isn't accepted on later models. If someone's tutorial tells you to set a thinking budget, it's out of date for the current lineup.
Model IDs and where they run
The API IDs are claude-fable-5-1, claude-opus-5-5, claude-sonnet-5-5 and claude-haiku-5-5. The same IDs work on the Claude API, Claude Models on Google Cloud, Microsoft Foundry and Claude Platform on AWS. Amazon Bedrock adds an anthropic. prefix, so Opus becomes anthropic.claude-opus-5-5. Every ID is a pinned snapshot, so the model behind it doesn't quietly change.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Retirement dates are Anthropic's commitment for the platforms it runs. Partner platforms set their own dates. Older models are still listed as available, including Claude Fable 5, Claude Opus 5, Claude Opus 4.8, Claude Sonnet 5, Claude Sonnet 4.6 and Claude Haiku 4.5. You'll also see Claude Mythos models on the pricing page marked limited availability, so don't go hunting for them in a normal account.
Which Claude model I'd start a small business on
I'd start on Opus 5.5, as Anthropic itself suggests, and only change once you've seen it work on your own tasks. Drop to Sonnet 5.5 when you want faster answers, and use Haiku 5.5 for high-volume jobs such as sorting enquiries or pulling details out of forms. Skip Fable 5.1 until Opus at higher effort fails your own test. For a deeper look at the default pick, see my guide to using Claude Opus 5.5; if you're torn between the top two, my Opus versus Sonnet comparison goes further, and my Sonnet versus Haiku page covers the fast end. Writers should read my pick of the best Claude model for writing.
If you're deciding which AI model your team should use for which job, that's the kind of AI enablement I help with. I run AI training and workshops for teams, and help businesses with AI enablement and workflow automation, so the busywork runs itself and people get time back. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.
Frequently asked questions
What are the current Claude models?
As of Nov 2026 Anthropic's docs list Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5 and Claude Haiku 5.5 as the current lineup.
Which Claude model should I use?
Anthropic says to start with Claude Opus 5.5 for most workloads, and to use Claude Fable 5.1 for demanding reasoning or when Opus 5.5 at higher effort still falls short.
How much do Claude models cost?
As of Nov 2026 base prices per million tokens are $10 input and $50 output for Fable 5.1, $4 and $20 for Opus 5.5, $2 and $10 for Sonnet 5.5, and from $0.10 and $0.50 for Haiku 5.5.
What is the context window of Claude models?
Every current model has a 1M token context window and 128K tokens of max output. Anthropic says 1M tokens is roughly 555k words on the current tokenizer.
When will Claude Opus 5.5 be retired?
Anthropic's docs say not sooner than 22 September 2027, for the platforms Anthropic operates. Partner platforms set their own dates.
