Anthropic released three generally available Claude models in September 2026: Claude Fable 5.1 on 1 September, Claude Opus 5.5 on 22 September and Claude Sonnet 5.5 on 28 September. Add Claude Haiku 4.5, and you have four models at very different prices. Here’s what each one is for, and how we’d choose for a business application.
The current Claude lineup
| Model | Best for | Price per 1M tokens (input / output) | Context window |
|---|---|---|---|
| Claude Fable 5.1 | Demanding reasoning and long-horizon agentic work | $10 / $50 | 1M tokens |
| Claude Opus 5.5 | Long-running agentic coding and knowledge work | $4 / $20 | 1M tokens |
| Claude Sonnet 5.5 | The best combination of speed and intelligence | $2 / $10 | 1M tokens |
| Claude Haiku 4.5 | The fastest and lowest-cost option, for simple high-volume work | $1 / $5 | 200K tokens |
The context window is how much text the model can take in at once. A million tokens is enough to work with long contracts, full manuals or a large codebase in a single request.
What’s new
Sonnet 5.5 is the new everyday model. Anthropic calls it “a clear upgrade over Sonnet 5 that runs 30% faster and costs up to 30% less for most work”, and positions it as “a faster, lower-cost complement to Opus 5.5” for well-scoped everyday tasks. The price per token is the same as Sonnet 5, so the saving comes from needing fewer tokens and steps for the same job. Early customers report the same: Slack says it beat Sonnet 5 on almost all of its offline Slackbot evaluations, in fewer steps and with about 14% fewer output tokens.
Opus 5.5 is for long, hard work. Anthropic positions it for long-running agentic coding and knowledge work. Its reasoning (“thinking”) is always on, and you control how hard it works with an effort setting, which defaults to medium.
Fable 5.1 is for the hardest jobs. It’s Anthropic’s most capable generally available model. Anthropic lists long-running agentic coding, multistep research, and document, spreadsheet and slide work, for jobs where Opus 5.5, even at higher effort, still falls short in your tests. It’s slower than Opus 5.5 and costs 2.5 times as much.
Fable or Mythos?
You may have seen Claude Mythos in the news. Claude Mythos 5.1 has the same capabilities and pricing as Fable 5.1, but only participants in Anthropic’s Project Glasswing can use it. Anthropic released Fable with strong safeguards to make it safer for general use. For almost every business, Fable 5.1 is the one you can actually use.
The June 2026 suspension, and what it teaches
On 12 June 2026, the US government applied export controls to Claude Fable 5 and Claude Mythos 5. Anthropic suspended access to both because it had no reliable way to verify users’ nationality in real time. The controls were lifted on 30 June, and Fable 5 came back for customers worldwide from 1 July. Mythos 5 was restored only for a set of US organisations.
The lesson isn’t about one company. Any model can be withdrawn, restricted or changed, whether by regulation, by the vendor or by an outage. Build your software so it can fall back to a second model, and keep the prompts and test cases that let you switch quickly.
Where Claude fits in business software
- Coding and engineering work: Opus 5.5 and Fable 5.1 are positioned for long-running agentic coding; Sonnet 5.5 handles everyday coding faster and cheaper.
- Documents, spreadsheets and slides: Fable 5.1 is listed for this work by name.
- Customer-facing assistants: Sonnet 5.5 for good answers at speed; Haiku 4.5 when volume is high and questions are simple.
- Long documents: Fable, Opus and Sonnet all take up to 1M tokens of context.
Where you can use Claude
- The Claude apps on web, iOS and Android, plus Claude Code for software teams.
- The Claude API, directly from Anthropic.
- Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry and Claude Platform on AWS.
If your company already has credits or a contract with AWS, Google Cloud or Microsoft, you can use Claude there instead of setting up a new vendor.
What it costs: a worked example
Take the same assistant as in our Gemini guide: 10,000 questions a month, about 2,000 tokens in and 500 tokens out each.
- Sonnet 5.5: 20 million input tokens × $2 = $40, plus 5 million output tokens × $10 = $50. About $90 a month.
- Haiku 4.5: 20 million × $1 = $20, plus 5 million × $5 = $25. About $45 a month.
Two features cut this further:
- Prompt caching. When every request starts with the same instructions and documents, the cached part costs $0.20 per million tokens on Sonnet 5.5 instead of $2, which is 90% less.
- Batch processing. Jobs that can wait, like overnight reports or bulk document tagging, cost 50% less.
How to pick
- Start with Sonnet 5.5.
- Use Haiku 4.5 for simple, high-volume steps.
- Move to Opus 5.5 for long, multi-step work where Sonnet’s answers aren’t good enough in your tests.
- Use Fable 5.1 only where Opus 5.5 at higher effort falls short, as Anthropic itself suggests.
- Keep a fallback model from another company ready.
FAQ
Which Claude model is best for coding?
Opus 5.5 is built for long-running agentic coding, and Fable 5.1 for the hardest work. Sonnet 5.5 handles everyday coding faster and for less. Test them on your own codebase before you decide.
What’s the difference between Claude Fable and Claude Mythos?
They have the same capabilities and price. Mythos is only for participants in Anthropic’s Project Glasswing; Fable has stronger safeguards and is generally available.
Is Claude Sonnet 5 still available?
Yes, as a legacy model. Anthropic says it won’t be retired before 30 June 2027 and recommends moving to Sonnet 5.5.
Can we use Claude through AWS or Google Cloud?
Yes: Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry and Claude Platform on AWS all offer it.
Planning an AI feature?
We build AI features into business software and test models on a sample of your real data before you commit. See our AI services or book a free 30-minute call.
Sources
- Claude Platform Docs: Claude Sonnet 5.5, Claude Opus 5.5, Claude Fable 5.1, Claude Sonnet 5
- Anthropic: Claude Sonnet, Introducing Claude Opus 5.5, Introducing Claude Sonnet 5.5
- Claude Platform Docs: Batch processing
- Anthropic: Redeploying Claude Fable 5 (30 June 2026)