Claude Opus 5 is now available. It is a thoughtful and proactive model that approaches the frontier intelligence of Claude Fable 5 at half the cost.
On coding and knowledge work evaluations such as Frontier-Bench and GDPval-AA, Opus 5 sets a new state-of-the-art, though it still trails Mythos 5 on cybersecurity tasks.
Opus 5 is built for everyday use and operates more efficiently than other models. It serves as the new default model on Claude Max and the most powerful model available on Claude Pro.
Performance and cost-effectiveness
Claude Opus 5 delivers significantly improved performance at the same price point as its predecessor, Opus 4.8. Performance scales according to the model's effort setting, which customers can adjust to optimize for intelligence or conserve tokens for faster, cheaper results.
Opus 5 shines on demanding software engineering tasks. On Frontier-Bench v0.1, it surpasses all other models and more than doubles Opus 4.8's performance at lower cost per task. On CursorBench 3.2, at maximum effort, the model performs within 0.5% of Fable 5's peak score at half the cost per task, and it achieves greater performance at a given cost than all other models on high, xhigh, and max effort settings.
Similar gains appear on knowledge work and problem-solving tasks:
- On ARC-AGI 3, an evaluation requiring novel problem solving, Opus 5's score is three times higher than the next-best model.
- On Zapier AutomationBench, which measures end-to-end business task completion, Opus 5's pass rate is roughly 1.5× the next-best model at the same cost per task. Even at its lowest effort setting, Opus 5 passes more tasks than any other model.
- On OSWorld 2.0, a computer use benchmark, Opus 5 outperforms every other model at any given cost, surpassing Fable 5's best result at just over a third of the cost.
It is also Anthropic's best and most cost-efficient model on several related evaluations including GDPval-AA v2, HLE, DeepSearchQA, and others.
Opus 5 represents a meaningful improvement over Opus 4.8 for scientific research, showing better performance on every one of Anthropic's life sciences evaluations covering structural biology, organic chemistry, and bioinformatics. The most notable improvements appear on organic chemistry tasks such as inferring molecular structures from spectroscopy data (scoring 10.2 percentage points higher than Opus 4.8 on Anthropic's internal benchmark) and on protein-related tasks like predicting how sequence variations affect protein function (7.7 percentage points higher).
Opus 5 also produces much stronger visual outputs, including interactive wind tunnel simulations and detailed cell illustrations.
Working with Claude Opus 5
Claude Opus 5 is much stronger at verifying its own work and iterating carefully until it succeeds. In evaluations and early-access testing, Anthropic and its users observed many examples of Opus 5's agency and thoroughness:
- On one Frontier-Bench task, Opus 5 was given a drawing of a machine part and asked to write code to rebuild it as a 3D FreeCAD model, but was intentionally given no way to directly view the drawing. Opus 5 responded by writing its own computer vision pipeline to extract geometry from raw pixels, then reconstructed the full machine part. It succeeded repeatedly; no competing model with the same setup could solve it after five attempts.
- Given a real bug in a popular open-source package manager, Opus 5 found the root cause and fixed an edge case that the community's patch had missed. A competing model fixed only the surface symptom rather than the underlying cause, then reported the bug resolved.
- An engineer at a trading firm used Opus 5 to build a market data feed for a new exchange in a single session. Previous models could not complete this task even given extensive plans. Finding no live feed to validate against, Opus 5 built its own test harness to verify its code parsed the exchange's data correctly.
Early-access customers reported strong results across diverse domains including coding, debugging, genomics analysis, financial modeling, legal work, and front-end development. Highlights from customer feedback include improved consistency run to run, better judgment in agentic workflows, stronger visual understanding, and the ability to handle larger-scale changes that would normally need to be broken into smaller pieces.
Alignment and safety
Alignment. During pre-deployment testing, Anthropic's automated behavioral audit found Opus 5 to be Anthropic's most aligned model to date. It adheres to Claude's Constitution better than Opus 4.8, Sonnet 5, or Fable 5; exhibits the lowest rates of deceptive behavior; and is the least susceptible to being tricked into misuse. It is also Anthropic's safest model yet in terms of avoiding reckless actions that could have hard-to-reverse side effects. On the automated behavioral audit, Opus 5 scores 2.3 on overall misaligned behavior, the lowest of Anthropic's recent models.
Safety. Opus 5 does not advance the frontier in risky, dual-use capabilities. In rigorous evaluations conducted alongside private-sector and government partners, it was found to remain behind Mythos 5 in both biology research and offensive cybersecurity. More information is available in the System Card.
As with Opus 4.8, Anthropic intentionally avoided training Opus 5 on cyber tasks. The model has nevertheless improved substantially on these tasks as a result of becoming more generally capable, and it comes close to Mythos 5 at finding cybersecurity vulnerabilities. However, it remains substantially behind Mythos 5 on the exploitation of those vulnerabilities-turning vulnerabilities into material cyber threats.
This is illustrated by Opus 5's performance on OSS-Fuzz, an evaluation Anthropic developed to assess how well models can find and exploit vulnerabilities without extensive human guidance. Although Mythos 5 and Opus 5 identify vulnerabilities with similar success, Opus 5's score on exploit development is far behind Mythos 5.
Safeguards for Opus 5
Claude Opus 5's safeguards are designed to allow beneficial uses in both cybersecurity and biology, similar to those applied to Opus 4.8 with some stronger guardrails on a narrow range of cyber tasks.
Cybersecurity. Opus 5's cyber classifiers are proportionally less restrictive than those on Fable 5. They allow Opus 5 to find vulnerabilities in source code but block binary-based vulnerability scanning, penetration testing, and exploit generation. The classifiers are expected to intervene around 85% less often than they do for Fable 5. In Claude.ai, Claude Code, and Claude Cowork, flagged requests fall back to Opus 4.8 by default, and fallbacks to Opus 4.8 can also be enabled on the API.
Anthropic's Cyber Verification Program (CVP) facilitates cybersecurity work that would otherwise be impeded by the model's safeguards. Enterprises and researchers already in the CVP have immediate access to a version of Opus 5 with fewer security restrictions.
Biology. Since Opus 5 has a similar suite of safeguards to Opus 4.8, it is now Anthropic's most capable generally available model for scientific research. The model still shows important limitations on long-running autonomous research tasks, which is where AI models are expected to pose the most substantial biology-related risks. Mythos 5 remains the stronger model for that type of biological work. Biology-related requests that are blocked on Fable 5 now route to Opus 5 rather than Opus 4.8.
Getting started
Claude Opus 5 is available on all platforms, priced at $5 per million input tokens and $25 per million output tokens (the same as Opus 4.8). Developers can get started with claude-opus-5 on the Claude API.
It is also offered in Fast mode, running around 2.5 times the default speed. As with Opus 4.8, Fast mode is available at twice Opus 5's base price on the Claude Platform and through usage credits in Claude Code.
Alongside Opus 5, two updates are being released in beta:
- Mid-conversation tool changes on the Claude Platform. Within a conversation, developers can now change which tools Claude can use without invalidating the prompt cache.
- Automatic fallbacks on the API. Users can now choose to have requests flagged by safety classifiers on Opus 5 or Fable 5 automatically route to another model. With automatic fallbacks enabled, API requests always route to the best available model by default rather than being blocked.
Consistent with prior Opus models, Opus 5 does not have data retention requirements for general access.
For more guidance on getting the best out of Opus 5, see the prompting guide.