Claude Opus 5.5: Release Date, Price, Benchmarks, How to Use
Updated 2026-10-07
Short answer: Claude Opus 5.5 is Anthropic's current top Opus model, released on September 22, 2026. It is the first model of the Claude 5.5 family, Anthropic says it performs at the level of Claude Fable 5.1 on most work, and it costs about 40% less to run than Claude Opus 5 on typical workloads. API pricing is $4 per million input tokens and $20 per million output tokens. You can use it in the Claude apps, in Claude Code and through the Claude Platform API.
People type the name in several ways – "opus 5.5", "opus5.5", "claude opus 5.5" – they all mean the same model. This page collects what Anthropic has officially published about it, with dates, so you can decide whether to switch.
Claude Opus 5.5 at a glance
| Claude Opus 5.5 | |
|---|---|
| Maker | Anthropic |
| Released | September 22, 2026 |
| Family | Claude 5.5 (Sonnet 5.5 and Haiku 5.5 announced "in the coming weeks") |
| Input / output price | $4 / $20 per 1M tokens |
| Cache reads / cache writes | $0.20 / $5 per 1M tokens |
| Fast mode | Up to 2.5x speed, $8 / $40 per 1M tokens (Claude Code and Claude Platform) |
| Speed vs Opus 5 | Generates output more than 30% faster (Anthropic) |
All figures above come from Anthropic's launch page "Introducing Claude Opus 5.5" (September 22, 2026).
What is new compared with Opus 5
It costs less. Per-token prices dropped by 20% on input and output ($5 → $4 and $25 → $20 per million). Cache reads, which Anthropic says make up most of the cost of agentic and coding work, dropped from $0.50 to $0.20 per million tokens – 60% less. Because the model also uses fewer tokens per task, Anthropic's own tests put the total saving at about 40% on typical workloads at default settings.
It is faster. Anthropic reports that Opus 5.5 generates output more than 30% faster than Opus 5, and a separate fast mode runs up to 2.5 times faster at a higher price.
It writes more clearly. The launch post says early testers found its writing easier to follow, with the most important information first – a direct response to feedback about Opus 5.
Long jobs got better. The examples Anthropic gives are long, sprawling engineering tasks: a 200,000-line codebase audited and fixed in under three hours (Opus 5 took over 20 hours and 2.5 times as many tokens), and a translation of the load balancer HAProxy from C to Rust that passed nearly all of its regression tests.
Usage limits went up. Together with the price drop, Anthropic increased the five-hour usage limits on the Pro, Max, Team and seat-based Enterprise plans and gave subscribers a rate-limit reset they can save and use when they choose.
Claude Opus 5.5 pricing
| Price per 1M tokens | Opus 5.5 | Opus 5 |
|---|---|---|
| Input | $4 | $5 |
| Output | $20 | $25 |
| Cache reads | $0.20 | $0.50 |
| Cache writes | $5 | $6.25 |
If your bill is mostly cached context – long coding sessions, agents that re-read the same files – the cache-read price is the number that matters most. For short one-off prompts, the 20% cut on input and output is what you will notice. Real costs depend on how many tokens your tasks use, so run a small sample of your own work before and after switching.
Benchmarks Anthropic published
Anthropic compared Opus 5.5 with Claude Fable 5.1, Claude Opus 5, GPT-6 Astra and GPT-5.6 Sol. A selection of the results from the launch page:
| Benchmark | Opus 5.5 | Fable 5.1 | Opus 5 | GPT-6 Astra |
|---|---|---|---|---|
| Terminal-Bench 4.0 (agentic coding) | 66.4% | 55.8% | 52.3% | 57.9% |
| FrontierCode v1.1 (main) | 54.4% | 50.3% | 48.0% | 53.3% |
| CursorBench 4.0 | 57.8% | 51.8% | 46.6% | – |
| Humanity's Last Exam (with tools) | 67.7% | 65.6% | 63.6% | 57.2% |
| OSWorld 2.1 (computer use) | 81.8% | 80.7% | 74.0% | – |
Two caveats from Anthropic itself: at this level, benchmark margins are "a less reliable guide to real-world differences", and in some tests the safety safeguards handed certain cybersecurity and biology tasks to older models, which likely lowered Opus 5.5's scores. GPT figures were taken from OpenAI's own reports. Treat the table as a direction, not a verdict, and test on your own tasks.
Anthropic also stresses efficiency: on FrontierCode at its default effort, Opus 5.5 beat GPT-6 Astra's top score at roughly a fifth of the cost per task, and on CursorBench it beat GPT-5.6 Sol by 11 points at about a third of the cost.
How to use Claude Opus 5.5
- In the Claude apps – open Claude on the web, desktop or mobile and pick Opus 5.5 in the model menu (availability depends on your plan).
- In Claude Code – choose Opus 5.5 as the model; fast mode is available there too if you want speed over price.
- Through the API – call Opus 5.5 on the Claude Platform. Anthropic publishes a "what's new" page and a migration guide for Opus 5.5; read them before switching production traffic, because some request settings changed (for example, Anthropic notes that thinking can't be disabled on this model).
- Through cloud platforms – Anthropic models are also offered through major cloud providers; check your provider's model list for availability.
Safety notes from the launch
Opus 5.5 is Anthropic's first release since it called for "pacing the frontier". External evaluators tested it before release, and Anthropic says it scored best of any model so far on its automated behavioral audit. Because its biology and cybersecurity skills are comparable to Claude Mythos 5.1, it ships with safeguards similar to those on Fable 5.1; vetted organizations can apply to verification programs for biology research and cybersecurity work. Full details are in the Opus 5.5 System Card.
Should you switch?
- You use Opus 5 today: switching is the obvious test – lower prices, faster output, and higher scores on Anthropic's coding and knowledge-work benchmarks.
- You compare against OpenAI models: see our GPT-6.1 vs Claude Opus 5.5 comparison for prices and shared benchmarks.
- You are choosing between Opus and Sonnet: Sonnet 5.5 has been announced but was not yet released on launch day; our Opus vs Sonnet guide explains how the two tiers usually differ.
FAQ
When was Claude Opus 5.5 released?
Anthropic announced and released Claude Opus 5.5 on September 22, 2026.
How much does Opus 5.5 cost?
$4 per million input tokens and $20 per million output tokens through the API, with cache reads at $0.20 and cache writes at $5 per million. Fast mode costs $8 / $40.
Is Opus 5.5 better than Opus 5?
On Anthropic's published benchmarks it scores higher on every shared test in the launch table, and Anthropic says it costs about 40% less to run on typical workloads.
Is "opus5.5" the same as Claude Opus 5.5?
Yes. "opus5.5", "opus 5.5" and "claude opus 5.5" are spellings of the same model name.
Are Sonnet 5.5 and Haiku 5.5 out?
At the Opus 5.5 launch, Anthropic said Sonnet 5.5 and Haiku 5.5 would follow "in the coming weeks". Check Anthropic's news page for the current status.