Anthropic says Claude Opus 5 delivers much of the performance of its Claude Fable 5 frontier model at half the cost per task, with configurable effort settings and availability through Claude subscriptions, the Claude Platform and Amazon Bedrock.
Anthropic has introduced Claude Opus 5, a high-end AI model that the company says comes close to the capabilities of its Claude Fable 5 frontier model while costing half as much per task.
According to Anthropic’s product announcement, Opus 5 is available immediately as the default model for Claude Max subscribers and the highest-capability model offered to Claude Pro users. The company is positioning the release around a balance of advanced performance and efficiency for recurring coding, analysis and knowledge-work tasks.
Anthropic said Claude Opus 5 sets a new high mark on the coding and knowledge-work evaluations it cited, including Frontier-Bench and GDPval-AA. The company also noted that Opus 5 remains behind Mythos 5 on cybersecurity tasks.
On Frontier-Bench v0.1, Anthropic said Opus 5 more than doubled the performance of Claude Opus 4.8 while reducing the cost per task. The company also reported that, at its maximum effort setting, Opus 5 scored within 0.5% of Claude Fable 5’s peak result on CursorBench 3.2 while operating at half the cost per task.
These are vendor-reported results rather than independent comparisons. Still, the figures illustrate Anthropic’s central argument for the release: customers may be able to use a less expensive model for demanding work that might otherwise require a frontier-tier system.
Anthropic said users can select an effort setting that changes the trade-off between capability, token consumption, latency and cost. The company described Opus 5 as providing improved performance at the same price as Opus 4.8, with higher-effort modes intended for more complex requests.
That design allows a single model to be used differently across workloads. Lower-effort settings are intended to conserve tokens and return answers more quickly, while high, extra-high and maximum effort settings are aimed at tasks where additional reasoning may be worth the added time and compute expense.
Anthropic’s announcement places particular emphasis on software engineering. It said Opus 5 performs strongly on valuable coding tasks and offers better performance at a given cost than competing models across its high, extra-high and maximum effort settings. The company also highlighted knowledge work and novel problem-solving evaluations as evidence of broader usefulness beyond programming.
Anthropic’s Claude Platform documentation lists the API model identifier as claude-opus-5. The documentation also specifies a one-million-token context window and describes default thinking behavior, pricing and availability.
AWS separately confirmed in a blog post that Claude Opus 5 is available through Amazon Bedrock and the Claude Platform on AWS. AWS described the model as Anthropic’s first fifth-generation Opus release.
The launch extends Anthropic’s effort to make model selection less dependent on choosing between a low-cost model and a frontier model. Opus 5 is presented as an option for organizations seeking advanced coding and professional-task performance while controlling per-task spending—an assertion that will ultimately depend on results in customers’ own workloads.
Anthropic launches Claude Opus 5 Anthropic has introduced Claude Opus 5, a high end AI model that the company says comes close to the capabilities of its Claude Fable 5 frontier model while costing half as much per task.
According to Anthropic’s product announcement, Opus 5 is available immediately as the default model for Claude Max subscribers and the highest capability model offered to Claude Pro users.
The company is positioning the release around a balance of advanced performance and efficiency for recurring coding, analysis and knowledge work tasks.
Continue reading
