Overview
Claude Opus 5 is a new generation flagship large model released by Anthropic on July 24, 2026, with intelligence level close to Claude Fable 5 but at half the price. It outperforms Opus 4.8 by more than twice on Frontier-Bench v0.1, and scores three times that of the next best model on ARC-AGI 3. Opus 5 becomes the default model for Claude Max and the strongest model for Claude Pro effective immediately, serving as the direct successor to Opus 4.8.
Key Features
- Leap in Performance: Frontier-Bench v0.1 performance more than double that of Opus 4.8, ARC-AGI 3 score three times the next best model; achieves new SOTA on GDPval-AA and CursorBench 3.2
- Half the Price: Intelligence level close to Fable 5 but at only half the price, significantly improved cost-effectiveness
- Default Model for Claude Max: Becomes the default model for Max and the strongest model for Pro effective immediately, available directly to subscribers
- Continuation of Agent and Coding Capabilities: Continues the Opus series' strengths in multi-step reasoning, autonomous Agent operations, and complex programming
- 1M Token Context: Maintains long-context capability, suitable for full-codebase and long-document analysis
- Effort Setting Feature: Supports adjustable effort settings, allowing users to optimize intelligence level or save tokens based on needs, achieving faster and cheaper results
Use Cases
- Complex software engineering tasks (architecture design / large-scale refactoring)
- Multi-step reasoning in scientific research, legal, and financial analysis
- Autonomous Agent scenarios (computer use / long-chain tool invocation)
- Enterprise scenarios requiring high model reliability
Pros
- Performance doubled compared to previous generation, industry-leading level
- Price halved, significantly improved cost-effectiveness
- Becomes default model for Max, available to subscribers immediately
- Agent and coding capabilities continue the Opus series' strengths
- Supports Effort Setting, flexible balance between intelligence and cost
Pricing
Opus 5 becomes the default model for Claude Max, and Pro users can access the strongest model. API price is halved compared to Opus 4.8, subject to Anthropic's official website. Available on Claude.ai Pro at $20/month and higher tiers for Max.
Summary
Claude Opus 5 is Anthropic's flagship leap in mid-2026, with doubled performance and halved price, directly becoming the default model for Max. If you are using Opus 4.8, upgrading to Opus 5 is almost a no-brainer: stronger and cheaper. Suitable for enterprise users with the highest model requirements, Agent developers, and complex programming scenarios.
Version History
- Opus 5 Official Release (2026-07-24): Performance more than double that of Opus 4.8, price halved; significant lead on Frontier-Bench v0.1, GDPval-AA, and CursorBench 3.2; lags behind on Mythos 5 cybersecurity tasks; new Effort Setting feature added; becomes default model for Claude Max and strongest model for Pro