Anthropic released Claude Opus 5 on July 24, 2026, positioning it as the company's new flagship for coding, agentic work, and long-running professional tasks. The model is now the default on Claude Max and the strongest available on Claude Pro, priced at $5 per million input tokens and $25 per million output tokens, unchanged from Opus 4.8.
Anthropic frames Opus 5 as coming close to the frontier intelligence of Claude Fable 5, the company's top-tier model, at half the price. At its maximum effort setting, Opus 5 lands within 0.5% of Fable 5's peak score on CursorBench 3.2, a coding benchmark, while costing half as much per task. On OSWorld 2.0, a test of computer-use agents, it surpasses Fable 5's best result at just over a third of the cost.
Anthropic says Opus 5 more than doubles Opus 4.8's score on Frontier-Bench v0.1, a software-engineering evaluation Anthropic ran internally, and scores three times as high as the next-best model on ARC-AGI 3, a test of reasoning on novel problems. On Zapier's AutomationBench, a workflow-automation benchmark, Opus 5's pass rate runs about 1.5 times the next-best model at the same cost per task.
The model ships with a 1-million-token context window, now both the default and the maximum, with no smaller variant, and a 128,000-token output limit. Thinking runs on by default, a change from Opus 4.8, where it had to be explicitly switched on, and users now set how hard the model works via a five-level effort scale from low to max. A Fast mode research preview runs Opus 5 about 2.5 times faster on the Claude API, at $10 per million input tokens and $50 per million output tokens, double the standard rate, and not yet available on Amazon Bedrock, Google Cloud, or Microsoft Foundry.
Opus 5 is live on the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry, alongside claude.ai, Claude Code, and Cowork. It replaces Opus 4.8 as the top-of-lineup option for Max subscribers just over three weeks after Anthropic made Claude Sonnet 5 the default model for Free and Pro users at an introductory $2 and $10 per million tokens, splitting the current lineup between a cheap default tier and a premium one built for heavier agentic work.
The launch lands eight weeks after Anthropic closed a $65 billion Series H at a $965 billion valuation, passing OpenAI to become the most valuable AI startup, and confidentially filed paperwork for an IPO. Opus 5's pricing keeps it level with Opus 4.8 rather than the $10-and-$50-per-million-token tier Anthropic set for Fable 5 and Mythos 5 in June, the two-tier split that Washington briefly forced offline over export-control concerns before clearing it for trusted partners.
Separately, Claude Code Desktop's iOS Simulator support entered public beta on July 21, three days before the Opus 5 launch, according to MacRumors. The feature adds a live simulator pane next to the Claude Code conversation on Mac, letting developers watch Claude build, run, and test iOS apps without needing macOS Accessibility or Screen Recording permissions. Per Anthropic's own documentation, it requires Claude Desktop version 1.24012.0 or later and Xcode with the iOS platform installed, and works in local sessions only, since cloud and SSH sessions run on a machine that can't reach simulators on the user's Mac.
Source: Anthropic, MacRumors, Claude Code documentation.
Comments