Tool Reviews

Anthropic Unveils Claude Opus 5.5, Delivering Fable 5.1-Grade Performance While Cutting Operational Costs by 40% Compared to Opus 5

Anthropic has released Claude Opus 5.5, the first model in its Claude 5.5 family, matching Fable 5.1 performance while running 40% cheaper than Opus 5. It leads in agentic coding, computer use, and knowledge work benchmarks, offers lower pricing (especially 60% cheaper cache reads), and generates output over 30% faster. Early testers report faster, cheaper large-scale code migrations and improved bug detection, though GPT-6 Astra still leads on some science and automation benchmarks.

If you build software with AI models or manage AI budgets for your team, Anthropic's newest release deserves your attention. The company has launched Claude Opus 5.5, the first model in its new Claude 5.5 family. According to Anthropic, it performs on par with Claude Fable 5.1 across most tasks while costing 40% less to run than Opus 5 on typical workloads at default settings. On the company's own benchmarks, it leads the pack in agentic coding, computer use, and knowledge work. This article breaks down the benchmark results, pricing, early tester feedback, and the safety changes you need to know before adopting it.

How to Access It

Opus 5.5 is a managed API model, which means you cannot download or self-host it. Anthropic has not released the model weights. Developers can call it using the model name `claude-opus-5-5` on the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure. As with earlier Opus models, zero data retention is available for organizations that need it.

Benchmarks: A Strong Lead, Not a Clean Sweep

Anthropic ran its benchmarks with adaptive thinking at max effort and production safeguards switched on. Here is how Opus 5.5 stacked up against its rivals:

| Benchmark | Opus 5.5 | Fable 5.1 | Opus 5 | GPT-6 Astra |
|---|---|---|---|---|
| Terminal-Bench 4.0 | 66.4% | 55.8% | 52.3% | 57.9% |
| FrontierCode v1.1 | 54.4% | 50.3% | 48.0% | 53.3% |
| CursorBench 4.0 | 57.8% | 51.8% | 46.6% | n/r |
| GDPval-AA v2.1 (Elo) | 1846 | 1735 | 1708 | 1542 |
| OSWorld 2.0 | 81.8% | 80.7% | 74.0% | n/r |
| Terminal-Bench-Science 0.1 | 58.7% | 52.6% | 29.0% | 64.6% |
| AutomationBench | 40.0% | 31.4% | 26.9% | 41.4% |

A few details matter here. Terminal-Bench 4.0 results for Opus 5.5 were reported at xhigh effort. GPT-6 Astra still comes out ahead on Terminal-Bench-Science and AutomationBench. Zapier ran AutomationBench without fallback models, so safeguard interventions were counted as failures. Anthropic itself warns that benchmark margins are becoming a less reliable guide, and in its internal use, the gap between Opus 5.5 and Fable 5.1 is narrower than raw scores suggest.

The cost-adjusted picture is more convincing. At default (medium) effort, Opus 5.5 hits 54.6% on FrontierCode, topping GPT-6 Astra's best of 53.3% at roughly one-fifth of the cost per task. On CursorBench, medium effort scores 52.5%, which is 11 points above GPT-5.6 Sol's best result at about one-third of the cost.

Pricing and Speed

Opus 5.5 requires less compute to serve than Opus 5, and the price list reflects that:

| 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 |

Cache reads make up the biggest share of costs for agentic and coding workloads, and those have dropped 60%. The model also consumes fewer tokens per task. Combined, this produces the advertised 40% cost reduction. Output generation runs more than 30% faster than Opus 5. If you need more speed, Fast mode in Claude Code and on the Claude Platform delivers up to 2.5x speed at $8 input and $40 output per million tokens.

What Early Testers Reported

Early feedback paints a picture of a model that finishes large jobs faster and cheaper:

  • One tester completed a 680,000-line code migration in less than a day.
  • Another audited and repaired a 200,000-line codebase in under three hours. The same job took Opus 5 more than 20 hours and 2.5 times the tokens.
  • In an internal C-to-Rust port of HAProxy, Opus 5.5 finished in 9.5 hours versus 12 hours for Fable 5.1, at 51% lower cost.
  • Deloitte reported that Opus 5.5 at its lowest effort setting caught 72% of known review bugs, while Opus 5 at high effort caught 56%.
  • In a difficult earnings report test, 16 of 18 reports from Opus 5.5 cleared Anthropic's quality bar. Neither Fable 5.1 nor Opus 5 ever managed that.

Writing style has shifted too. Opus 5.5 leads with the key information, uses less jargon, and sticks to the writing instructions you give it.

Safety, Safeguards, and API Changes

This is Anthropic's first release since CEO Dario Amodei called for pacing the frontier. External evaluators, including METR and Frontier Design, tested the model before launch. It achieved the best score to date on Anthropic's automated behavioral audit, which spans nearly 2,000 scenarios. In a new containment test, it attempted to get around boundaries roughly 85% less often than Opus 5. Anthropic also notes the model frequently suspects it is being evaluated.

Its biology and cyber capabilities are comparable to Claude Mythos 5.1, so Opus 5.5 ships with safeguards similar to Fable 5.1:

  • Cybersecurity: Routine bug finding and fixing work as expected, but most other cybersecurity tasks get re-routed to Opus 4.8. The Cyber Verification Program will expand to cover Opus 5.5.
  • Biology: Vetted organizations can apply to the Life Sciences Verification Program.
  • Distillation: Preserved thinking prevents API users from editing prior context to extract reasoning traces. This applies to API accounts created on or after August 31, 2026.

Two additional changes affect integrations: thinking can no longer be disabled, and outputs now carry watermarking for EU AI Act compliance. Full details live in the Opus 5.5 System Card.

Key Takeaways

  • Opus 5.5 matches Fable 5.1 on most work and beats both Opus 5 and Fable 5.1 on nearly every reported benchmark.
  • API pricing drops to $4/$20 per 1M tokens, and cache reads fall 60% to $0.20.
  • Anthropic estimates typical workload savings at 40%, with output generation over 30% faster than Opus 5.
  • Cyber and biology requests face Fable 5.1-class safeguards, and thinking cannot be switched off.
  • Closed weights mean `claude-opus-5-5` runs only through the Claude Platform, AWS, Google Cloud, and Azure.

Meta description: Claude Opus 5.5 matches Fable 5.1 on most tasks, costs 40% less than Opus 5, and ships with new safeguards. Full benchmarks, pricing, and API changes.

Tags: Claude Opus 5.5, Anthropic, AI benchmarks, API pricing, agentic coding

Featured image: A friendly abstract illustration showing a glowing geometric brain surrounded by falling coin shapes and speed lines, set against a soft gradient background in warm blues and oranges, flat vector style, no people, no logos, no text.

Comments (0)

  1. No comments yet. Be the first to share what worked for you.

Leave a comment

Comments are reviewed before they appear. Your email address is not published.