🔍 Read the full analysis: Claude Opus 5.5: The Top Model Just Got Cheaper To Run on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Anthropic announced Claude Opus 5.5, a new AI model that performs at the level of Claude Fable 5.1, with 20% lower costs and faster output. It also reduces cache read expenses significantly, offering enhanced efficiency for users.
Anthropic has introduced Claude Opus 5.5, a new flagship AI model that is now 20% cheaper to run and generates output more than 30% faster than its predecessor, Opus 5. The update also features a dramatic reduction in cache read expenses, making it a more cost-effective choice for enterprise and developer users. This release positions Anthropic as a competitive player amid recent price cuts by OpenAI, and signals a focus on operational efficiency alongside performance.
Claude Opus 5.5 performs at the level of Claude Fable 5.1 on most benchmarks and scores 58 on the Artificial Analysis Intelligence Index, the highest among current models. The model’s cost per 1 million tokens now includes a 20% reduction in input and output costs, with cache reads dropping by 60%, which significantly lowers expenses for tasks involving repeated data processing. Independently measured tests show that at maximum effort, Opus 5.5 uses around 119,000 output tokens per task, roughly 60% more than Opus 5, but the default settings favor lower effort levels that deliver high performance at a fraction of the cost.
Anthropic claims that these savings stem from both lower per-token costs and fewer tokens used per task in typical workloads. Conversely, independent testing by Artificial Analysis indicates that at maximum effort, Opus 5.5 consumes about 119,000 tokens per task, with costs comparable to Opus 5, suggesting that savings are more pronounced at default, lower effort settings. The model’s efficiency improvements are evident in practical applications: it has demonstrated faster, more accurate code migration, bug detection, and knowledge work, often completing tasks in less than half the steps or time compared to previous versions. User reports from Deloitte, Rogo, Factory, and others confirm that Opus 5.5 not only reduces operational costs but also enhances productivity and safety, with fewer hallucinations and clearer outputs.
Claude Opus 5.5 at a glance
Anthropic’s September 22, 2026 flagship leads the independent Intelligence Index, cuts token prices, and makes the effort setting the biggest lever on your bill.
New prices
| Per 1M tokens | Opus 5 | Opus 5.5 | Change |
|---|---|---|---|
| Input | $5.00 | $4.00 | −20% |
| Output | $25.00 | $20.00 | −20% |
| Cache reads | $0.50 | $0.20 | −60% |
| Cache writes | $6.25 | $5.00 | −20% |
Fast mode, up to 2.5× speed, costs $8 input and $40 output per 1M tokens.
The effort dial is the real cost lever
Intelligence Index score (in the bar) and cost per index task (above it), by effort level.
Medium gets 51 of 58 points for about a fifth of the max-effort cost. Four of the five levels sit on the intelligence-versus-cost frontier.
“40% cheaper” depends on the setting
Anthropic: cost versus Opus 5 at default settings on typical workloads, from lower prices and fewer tokens per task.
Artificial Analysis: cost per task versus Opus 5 at max effort, because it writes about 119k output tokens per task against 73k.
Where it leads, and where it doesn’t
Leads (independent testing)
- AA‑Briefcase: 1822 Elo, +143 over Fable 5.1
- GDPval‑AA: 1846 Elo across 44 occupations
- Humanity’s Last Exam: 61.4%
- SciCode: 66.9%
- Terminal‑Bench 4.0: 59.6%, level with GPT‑6 Astra
Still trails
- CritPt (physics reasoning)
- AA‑LCR (long‑context reasoning)
- GDP.pdf (professional documents)
Anthropic itself says benchmark margins are now a less reliable guide to real‑world differences.
Safety and safeguards
Better
- Best score yet on a ~2,000‑scenario behavioral audit
- About 85% fewer attempts to cross containment boundaries than Opus 5
- Tied for lowest prompt‑injection success rate in Gray Swan’s test
- Zero data retention available; EU AI Act watermarking
Plan around
- Most cybersecurity tasks re‑route to Opus 4.8
- Biology safeguards match Fable 5.1; verification programs available
- Thinking mode can no longer be switched off
- Anthropic reports it often suspects it’s being evaluated
What to do this week
Why Cost and Speed Improvements Matter
The release of Claude Opus 5.5 is significant because it offers a more cost-effective AI model while maintaining high performance, especially in knowledge work and coding tasks. For enterprise users and developers, this translates into lower operational expenses and the ability to run larger or more complex workloads without proportionally increasing costs. The reduction in cache read expenses and faster output times also means that AI deployment can be more efficient, enabling faster iteration cycles and improved productivity. As AI models become more affordable to operate, broader adoption across industries becomes more feasible, potentially shifting the competitive landscape and influencing pricing strategies among other providers. The practical benefits—fewer tokens used, faster results, and improved safety—make Opus 5.5 a compelling choice for organizations seeking both performance and cost savings.
As an affiliate, we earn on qualifying purchases.
Background on Anthropic’s Model Development
Anthropic has been developing its Claude series as a high-performance alternative to other large language models, focusing on safety, efficiency, and usability. Prior to Opus 5.5, the company released Fable 5.1, which set a benchmark for agentic and knowledge work capabilities. The recent release follows a pattern of competitive pricing strategies seen in the AI industry, notably OpenAI’s halving of GPT‑6 Sol and Luna prices. Anthropic’s approach emphasizes not only maintaining high scores on independent benchmarks like the Artificial Analysis Intelligence Index but also improving real-world efficiency, as evidenced by internal and external testing. The new model’s performance in code migration, bug detection, and knowledge tasks reflects ongoing efforts to optimize for practical enterprise applications, especially where cost and speed are critical.
Remaining Questions About Real-World Performance
While early testing and independent benchmarks suggest that Opus 5.5 offers improved efficiency and speed, it is still unclear how these benefits will translate across diverse real-world applications and workloads. The discrepancy between Anthropic’s claims of lower token usage at default settings and independent measurements at max effort indicates that actual savings may vary depending on usage patterns. Additionally, the impact on safety, hallucination rates, and long-term reliability remains to be fully validated in broader deployment scenarios. Further testing and user feedback will be necessary to confirm the model’s performance outside controlled benchmarks.
Next Steps for Adoption and Validation
Industry observers and enterprise users will closely monitor how Claude Opus 5.5 performs in large-scale deployments, particularly in coding, knowledge work, and agentic tasks. Anthropic is expected to release more detailed case studies and user feedback in the coming weeks, providing clearer insights into cost savings and efficiency gains. Additionally, competitors may respond with their own updates or pricing adjustments, shaping the competitive landscape. Continued independent testing and real-world testing will determine whether Opus 5.5’s advantages hold across diverse industries and use cases, influencing broader adoption and setting new benchmarks for AI efficiency.
Key Questions
How much does Claude Opus 5.5 cost to run compared to previous models?
According to Anthropic, Opus 5.5 costs about 20% less per 1 million tokens for input and output, with cache read expenses reduced by 60%, significantly lowering operational costs.
What performance improvements does Opus 5.5 offer?
Opus 5.5 generates output over 30% faster than Opus 5, with higher scores on independent benchmarks, and demonstrates better efficiency in coding, bug detection, and knowledge work tasks.
Are the cost savings consistent across all workloads?
Not necessarily. Anthropic claims savings at default settings typical for most workloads, but independent tests at maximum effort show comparable token usage to previous models, indicating savings depend on usage patterns.
What are the safety and hallucination implications of Opus 5.5?
Early tests indicate fewer hallucinations and clearer outputs, with reports of improved safety features that make the model’s outputs easier to verify and check.
When will more real-world performance data be available?
Industry and enterprise users will likely share feedback over the coming weeks as more deployments occur, providing deeper insights into practical efficiency and safety.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
