Claude Opus 4.5 Crushes Competition in Coding Performance Tests

TLDR

• Claude Opus 4.5 launched November 24, 2025, as Anthropic’s most advanced AI model to date
• The model outscored every human candidate on Anthropic’s performance engineering exam within a two-hour limit
• Pricing starts at $5 per million input tokens and $25 per million output tokens
• New features include expanded Chrome integration and Excel functionality for enterprise users
• The release follows Microsoft and Nvidia investments that valued Anthropic at $350 billion

Anthropic unveiled Claude Opus 4.5 on Monday, introducing its most capable artificial intelligence model yet. The company released this model just weeks after launching Claude Sonnet 4.5 and Claude Haiku 4.5.

The startup was founded by former OpenAI employees in 2021. Last week, Microsoft and Nvidia invested billions in the company, pushing its valuation to approximately $350 billion.

Claude Opus 4.5 demonstrates advanced capabilities in software development and enterprise applications. The model targets professional developers, financial analysts, consultants, and accountants who need AI assistance with complex tasks.

Testing Against Human Performance

Anthropic tested Claude Opus 4.5 using the same take-home exam given to performance engineering job applicants. The test measures technical ability and decision-making under time pressure. Within the standard two-hour window, the AI model achieved a higher score than any human candidate in company history.

The model also topped SWE-bench Verified, an industry standard test for software engineering capabilities. Claude Opus 4.5 surpassed competing models from Google and OpenAI on this benchmark.

Internal testing teams at Anthropic reported consistent results. Testers found the model handles unclear requirements without extensive guidance. The AI identifies and fixes bugs across multiple systems independently.

Technical Capabilities and Features

Claude Opus 4.5 processes spreadsheets, presentations, and research tasks more effectively than earlier versions. The model uses fewer tokens to reach solutions compared to previous releases.

Anthropic introduced a new effort parameter in its API. This feature lets developers balance between speed and capability based on their specific needs. At medium effort settings, Claude Opus 4.5 matches earlier performance while using 76% fewer output tokens.

The company expanded its security measures for this release. Claude Opus 4.5 shows improved resistance to prompt injection attacks compared to other frontier models. These attacks attempt to trick AI systems into harmful behavior through deceptive instructions.

Product Availability and Pricing

Developers can access Claude Opus 4.5 through Anthropic’s API using the identifier claude-opus-4-5-20251101. The model is available on the company’s apps and three major cloud platforms.

Input tokens cost $5 per million while output tokens are priced at $25 per million. This pricing structure makes the technology accessible to a wider range of businesses and development teams.

Claude for Chrome now reaches all Max subscription users. The browser extension performs tasks across multiple open tabs. Claude for Excel became generally available to Max, Team, and Enterprise subscribers.

The company added Claude Code functionality to its desktop application. Users can now run multiple coding sessions simultaneously. The Plan Mode feature asks clarifying questions before building detailed execution plans.

Anthropic removed usage caps specific to Opus models for subscribers with access. Long conversations in the Claude interface continue without message limits through automatic context summarization.

The post Claude Opus 4.5 Crushes Competition in Coding Performance Tests appeared first on Blockonomi.