Claude Opus 5
Anthropic's Claude Opus model for coding, agents and knowledge work
Claude Opus 5 is a large language model from Anthropic, released as the newest member of the Claude Opus line. The company says it approaches the intelligence of Claude Fable 5 at half the price, with strong results on software engineering, automation and computer-use evaluations.
- More than doubles Opus 4.8's Frontier-Bench v0.1 score at lower cost per task
- Within 0.5% of Fable 5 on CursorBench 3.2 at max effort, at half the cost
- ARC-AGI 3 score reported at three times the next-best model
- Beats every model at any given cost on OSWorld 2.0 computer use
- Default model on Claude Max, strongest model on Claude Pro
What it is
Claude Opus 5 is a general-purpose language model made by Anthropic and the newest entry in its Claude Opus series. Anthropic describes it as thoughtful and proactive, aimed at everyday use in coding, knowledge work and long-running agent tasks. The company positions it just below Claude Fable 5 on raw intelligence while costing half as much per task. It is offered with an effort setting that lets users trade intelligence against token use, and it is the default model on the Claude Max plan and the strongest option on Claude Pro.
How it came about
Anthropic announced Claude Opus 5 as available immediately, succeeding Opus 4.8 in the Opus line. The company reports improved results at the same price as that predecessor. Early-access customers included Cursor, Zapier, Box, Lovable and others quoted in the announcement, who tested the model on coding, automation, finance, legal and research workflows before general release.
Inside the model
Anthropic has not published architectural details such as layer counts, parameter totals or training data for Opus 5. The announcement describes an effort setting that adjusts how much thinking the model does, letting customers optimise for intelligence or save tokens for faster, cheaper runs. The company emphasises efficiency: better results than Opus 4.8 at the same cost, and lower cost per task than comparable models at high effort levels. It also highlights stronger self-verification, careful iteration and the ability to build its own tooling, such as a computer vision pipeline and a test harness, when solving problems.
What it can do
Anthropic reports that Opus 5 leads on Frontier-Bench v0.1 and GDPval-AA, more than doubling Opus 4.8's Frontier-Bench result at lower cost per task. On CursorBench 3.2 at max effort it lands within 0.5% of Fable 5's peak score at half the cost. The company says its ARC-AGI 3 score is three times the next-best model, and that it beats every model at any given cost on OSWorld 2.0. It trails Mythos 5 on cybersecurity tasks. Life sciences results improved over Opus 4.8, notably organic chemistry and protein function prediction.
How to use it
Claude Opus 5 is available through Anthropic's Claude app and API. It is the default model for Claude Max subscribers and the strongest model offered on Claude Pro. Customers can select an effort setting to balance quality against token spend. No downloadable weights or open licence are described.
Safety and testing
Anthropic's announcement does not detail safety evaluations, red-teaming or incidents for Opus 5.
Reception and debate
Early-access users quoted by Anthropic describe Opus 5 as a clear step up from Opus 4.8, praising debugging, root-cause analysis, numerical reasoning, legal agent work and consistency across runs. Cursor reported it just under Fable 5 on CursorBench, and Zapier said it topped its AutomationBench leaderboard. Box measured 8% gains over Opus 4.8 overall, with larger improvements in data analysis and due diligence. These are customer statements reported by the company.
Why it matters
Anthropic frames Opus 5 as a model built for daily use rather than peak intelligence alone, pairing near-frontier results with lower cost per task. The company's examples show it handling multi-step engineering work, such as rebuilding a machine part from a drawing and fixing a missed edge case in an open-source package manager. Its position as the default on Claude Max puts it in front of a wide developer audience.
Questions people ask about Claude Opus 5
What is Claude Opus 5?
It is Anthropic's newest Claude Opus language model, aimed at coding, knowledge work and agent tasks. The company says it reaches close to the intelligence of Claude Fable 5 while costing half as much per task, and it is the default model on the Claude Max plan.
How is it different from Opus 4.8?
Anthropic reports much better performance at the same cost. On Frontier-Bench v0.1 it more than doubles the earlier model's result at lower cost per task, and it improves on every life sciences evaluation the company runs, with the largest gains in organic chemistry and protein-related work.
What is the effort setting?
It is a customer-facing control that changes how much thinking the model does. Higher settings push for more intelligence, while lower settings conserve tokens for faster and cheaper results. Anthropic's charts report performance separately for each effort level.
Where does it fall short?
Anthropic states that Opus 5 remains behind Mythos 5 on cybersecurity tasks. The announcement gives no parameter count, context window or training data details for the model.
How do I get it?
It is available through the Claude app and API. It is the default model for Claude Max subscribers and the strongest model on Claude Pro. Anthropic describes no downloadable weights or open licence for it.
Good reading and sources
There is no English Wikipedia article on Claude Opus 5 yet, so the facts on this page come from the sources listed above: anthropic.com/news/claude-opus-5. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryLLM wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.