Anthropic launches Claude Opus 4.6 with performance upgrades and coding accuracy

AI startup Anthropic continues to make headlines across the tech industry with a series of launches, from Cowork and AI agents to its latest model, Claude Opus 4.6. Just two months after releasing Opus 4.5, the company has rolled out Opus 4.6, pointing to it as the most advanced model for coding.

Claude Opus 4.6 reportedly outperforms its predecessor in coding skills, and offers improved performance in managing longer or multi-step coding, debugging, and it also claims to reduce errors. It comes with a 1-million-token context window in beta, offering greater reliability.

Anthropic’s press note states that Claude Opus 4.6 can also perform everyday tasks such as analysing financial reports, researching, generating documents, spreadsheets, presentations, and more. In addition, it is also integrated into Cowork, allowing Claude to multitask across workflows.

Anthropic claims that the Opus 4.6 upgrade will be better at working in teams and handling long, complex jobs. With Claude Code, users can set up multiple Claude agents that can perform specific tasks such as coding, reviewing and testing.

The company has also introduced advanced capabilities such as adaptive thinking and Claude in PowerPoint. In addition, it can also intelligently summarise what it’s already done.

Claude Opus 4.6 benchmark tests

In early benchmark tests, the Claude Opus 4.6 outperforms all other frontier AI models. In agentic coding evaluation (Terminal-Bench 2.0), multidisciplinary reasoning test, and knowledge work in finance, legal, and related domains, and BrowseComp, the Opus 4.6 achieves the highest scores in comparison to OpenAI’s GPT-5.2, Gemini 3 Pro, and Sonnet 4.5.

A common complaint about AI models is “context rot,” where performance degrades as conversations exceed a certain number of tokens. Anthropic claims that Opus 4.6 performs markedly better than its predecessors: on the 8-needle 1M variant of MRCR v2, a needle-in-a-haystack benchmark that tests a model’s ability to retrieve information “hidden” in vast amounts of text, Opus 4.6 scores 76%, whereas Sonnet 4.5 scores just 18.5%.

Claude Opus 4.6 availability 

Claude Opus 4.6 is available on claude.ai, API, and all major cloud platforms. If you’re a developer, use claude-opus-4-6 via the Claude API. Pricing for the subscription remains the same, which starts at $5.