Which Claude model should your team use?
In brief
The November 2026 lineup: Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5. How to pick, why Sonnet 5.5 is the right default for most work, and why tuning effort usually beats switching models.
Contents
Most teams spend too long picking a model and too little time setting it up. Here is the short version.
Last checked: November 22, 2026, against Anthropic's model and pricing pages. Prices are per million tokens (input / output) on the API.
The current lineup
| Model | Price (in / out) | Best for |
|---|---|---|
| Claude Fable 5.2 | $8 / $40 | The hardest reasoning and long-running agents |
| Claude Opus 5.5 | $4 / $20 | Complex coding, agentic work and high-stakes writing |
| Claude Sonnet 5.5 | $2 / $10 | Everyday work: most chat, coding and agents |
| Claude Haiku 5.5 | $0.10 / $0.50 | High-volume, fast, simple tasks |
All four have a 1M-token context window and up to 128K tokens of output. Haiku 5.5 charges more ($0.50 / $2.50) once a prompt passes 100K tokens. The other three charge the same rate across the whole window.
Older models are being retired. Sonnet 4.5 retires on November 30, 2026, and fast mode for Opus 4.8 goes away the same day. If anything you run still names an older model, move it now. Each week's changes are in What changed in Claude this week.
What each model is good at
Sonnet 5.5 is the workhorse. It is fast, strong at coding and tool use, and cheap enough to run all day. Start here.
Opus 5.5 is for work where a better answer is worth paying for: tricky codebases, long documents with real judgment in them, and multi-step agents. It is cheaper than Opus 5 was ($5 / $25).
Fable 5.2 is Anthropic's most capable widely available model. Use it for the hardest problems and for agents that run for a long time on their own. It thinks on every request and can take minutes on hard tasks, so it is not for quick chat.
Haiku 5.5 is now a real option, not just a budget pick. It has the same context window and thinking as Sonnet 5.5 at a twentieth of the price on short prompts. Use it for sorting, tagging, extraction and simple replies at scale. Test it on your task before you rule it out.
What about Claude Mythos 5.2?
Mythos 5.2 is the same model as Fable 5.2, at the same price, with fewer safeguards for cybersecurity and biology research. It is only for organizations verified through Anthropic's programs, like the Cyber Verification Program, and is available on the API and the big clouds, not as a choice in the Claude app. If you aren't a security team or a life-science lab, Fable 5.2 is the one you can use. (Anthropic's Mythos 5.2 page)
The mistake teams make
They pick the biggest model for everything "to be safe". That multiplies the bill and often makes simple work slower, without making it better.
The opposite mistake is just as common: dropping to the cheapest model to save money, then spending staff time fixing the output.
The lever most teams miss: effort
Every current model has an effort setting: low, medium, high, xhigh and max. It controls how much the model thinks before it answers.
- Low for chat, quick lookups and simple sub-tasks
- Medium for most everyday work
- High or xhigh for coding, agents and anything hard
- Max only when being right matters more than cost
Turning effort down on a strong model usually beats switching to a weaker one. Note the defaults differ: Sonnet 5.5 defaults to high, while Opus 5.5 and Haiku 5.5 default to medium. Set it on purpose rather than relying on the default.
For team admins: model strategy
- In Claude apps, people pick the model per chat. On Enterprise, admins can control which models and effort levels each role can use, and an Efficient mode keeps routine work on Sonnet and Haiku.
- In Claude Code, the default is set per user or per project. Most teams run Sonnet 5.5 or Opus 5.5 and save Fable 5.2 for the hardest jobs.
- On the API, pick a model per job, not per company. A support bot, a nightly report and a coding agent will each have a different best answer.
The simple rule
Default to Sonnet 5.5. Move up to Opus 5.5 or Fable 5.2 when the work is hard and the answer matters. Move down to Haiku 5.5 when the work is simple and the volume is high. Tune effort before you switch models.
Quick answers
Which Claude model is best for most work? Claude Sonnet 5.5. It is fast, strong at coding and tool use, and costs $2 / $10 per million tokens. Move up to Opus 5.5 or Fable 5.2 only when the work is hard and the answer matters.
What is the difference between Claude Fable 5.2 and Mythos 5.2? They are the same model at the same price. Mythos 5.2 has fewer safeguards for security and biology research and is only for organizations Anthropic has verified.
Is Claude Haiku 5.5 good enough for real work? For simple, high-volume tasks like sorting, tagging and extraction, often yes, at a twentieth of Sonnet's price on short prompts. Test it on your own task first.
Further reading
- Models overview, Anthropic's official list of models, prices and limits
- Claude pricing
- Adaptive thinking and effort, in the glossary