Claude Opus 5.5 explained: what changed, what it costs

Anthropic released Claude Opus 5.5 on 22 September 2026. It is priced below Opus 5 and aimed at long-running coding and knowledge work.

Opus 5.5 is the first model in Anthropic’s Claude 5.5 family. Anthropic says it performs at the level of Claude Fable 5.1 on most work and costs about 40 per cent less to run than Claude Opus 5 on typical workloads. If your agents or Claude Code seats run on Opus 5, this is the week to price a trial. It is not the week to switch production.

The pitch is simple enough. Per-token rates fall and cache reads drop by 60 per cent, which matters most for agents that reread the same context on every turn. The catch: thinking cannot be switched off, several Opus 5 API patterns now return errors, and the headline scores are Anthropic’s own.

Everything below is checked against Anthropic’s launch page, API overview, what’s new page, migration guide and pricing page.

Claude Opus 5.5 at a glance

• Best suited to: long-running agentic coding and research that already runs on Opus 5.
• API price: $4 input and $20 output per million tokens.
• Cache reads: $0.20 per million tokens, down from $0.50.
• Context: one million tokens, with up to 128,000 output tokens (300,000 on the Batch API with the output-300k-2026-03-24 beta header).
• Model ID: claude-opus-5-5 (Bedrock: anthropic.claude-opus-5-5).
• Default effort: medium, where Opus 5 defaulted to high.

What Is Claude Opus 5.5?

Opus 5.5 is Anthropic’s new Opus model, and it is not the top of the range. Fable 5.1 still sits above it at $10 input and $50 output, the same list price as Fable 5. If Anthropic’s parity claim holds, that premium gets hard to justify for ordinary professional work.

It takes text and images and returns text. Adaptive thinking is always on, and you steer its depth with the effort setting. Knowledge cutoff is June 2026, and Anthropic will not retire the model before 22 September 2027.

It is available on Claude Pro, Max, Team and Enterprise, the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS. Five-hour usage limits rise on Pro, Max, Team and seat-based Enterprise, and paid users get a rate-limit reset they can save. Sonnet 5.5 and Haiku 5.5 follow in the coming weeks. If you run Opus 5 today, do not wait for them.

How Much Does Opus 5.5 Cost Compared With Opus 5?

Per million tokens, from Anthropic’s pricing page, with cache writes at the five-minute rate:

If the table extends beyond the screen, scroll sideways to view all columns.

Token rates are 20 per cent below Opus 5. Cache reads are 60 per cent below. Anthropic puts the net effect at about 40 per cent cheaper on typical workloads, but your bill still depends on prompt size, cache-hit rate, output length and effort.

Effort deserves a second look. The default drops from high to medium, so part of the saving comes from the model thinking less. If you pin effort to high to match Opus 5, measure the cost again.

Fast mode costs $8 input and $40 output for up to 2.5 times the speed, and the pricing page lists it on the first-party Claude API only. The Batch API takes 50 per cent off input and output.

A dollar price list is not a UK invoice. Convert at your own rate and add VAT where it applies. Zero data retention is available, but review the processor terms under UK GDPR before client data leaves the building.

Does It Match Claude Fable 5.1?

Anthropic says yes, on most work. It also says the real gap is narrower than the scores suggest. Keep both sentences in view.

These are Anthropic’s launch figures, not independent results:

If the table extends beyond the screen, scroll sideways to view all columns.

Anthropic also claims output more than 30 per cent faster than Opus 5, a 680,000-line migration in less than a day, and a C-to-Rust port of HAProxy in 9.5 hours against 12 for Fable 5.1, at 51 per cent lower cost. Treat those anecdotes as a test list, not a forecast of your sprint.

Honest answer: if you reach for Fable because Opus 5 keeps failing the same long job, keep a Fable budget. If you use Opus 5 because you can afford to leave it running, try Opus 5.5 first. As we said of Grok 4.6, token price is not the real cost. A finished, checked result is.

What Breaks If You Migrate From Opus 5?

Chat users can change the model picker and carry on. API integrations need a proper check. Anthropic lists four breaking changes and one behaviour change:

1. Thinking cannot be disabled. A disabled or fixed-budget thinking setting returns a 400 error. Control depth with effort instead.
2. Forced tool use returns an error. Swap tool_choice values of any or a named tool for auto with strict tool use and a clear prompt instruction.
3. Thinking blocks are tied to the model and conversation. Opus 5.5 keeps reasoning carried over from Opus 5, but drops blocks from Fable, and most models cannot read its blocks.
4. computer_20251124 is rejected on the Claude API and Google Cloud. Move to computer_toolset_20260801. Bedrock keeps the old tool.
5. Text between tool calls now arrives in thinking blocks. Those blocks are empty by default, so progress updates go quiet unless you change the display setting.

Preserved thinking is the quieter risk. For API accounts created on or after 31 August 2026 it applies to Opus 5.5 and Fable 5.1, and Anthropic describes it as an anti-distillation control. Change the system prompt, tools or earlier messages after an Opus 5.5 thinking block and the request returns a 400 error by default. Keep conversations append-only, and test any workflow that rewrites history.

Biology and cybersecurity requests run under Fable-class safeguards. Some flagged biology work is completed by Opus 5, and some flagged cyber work by Opus 4.8. The migration guide also asks you to handle the expanded refusal reasons and configure a fallback. Anthropic names Frontier Design and METR as external testers. The Life Sciences Verification Program is open to vetted organisations, and Cyber Verification Program access expands in the coming weeks. Read the system card before promising a client anything in either field.

Opus 5.5 also carries Anthropic’s EU AI Act watermarking. A detector may estimate that Claude was involved. It will not prove authorship or ownership.

What Should UK Teams Test First?

Pick one Claude Code or research agent that runs on Opus 5 today. Give Opus 5.5 the same brief and the same repository or documents. Record tokens in, tokens out, steps and minutes of human rework. Try Fable 5.1 only if Opus 5.5 stalls.

Good first tests for a Manchester or Leeds team:

• a site or campaign audit that rereads the same brand rules on every turn
• a migration or refactor already running in Claude Code
• a research brief that must cite primary pages, not recap blogs

Treat what comes back like a pull request from a capable new starter. Review it, check the citations, and keep a named person accountable for anything that publishes.

Should Your Business Switch to Claude Opus 5.5?

For most stacks already on Opus 5, yes, after a measured trial. Compare tokens, steps and rework against Opus 5 on one agent, then move traffic in stages. Do not flip production on launch day.

Keep Fable 5.1 for long, costly jobs where Opus 5.5 has already failed and the budget exists. Stay on Sonnet 5 when the work is short. If GPT-6 Astra is already in the stack, run it on the same task. No model wins every job.

If your team wants a measured trial plan, talk to AIWIZ.