Anthropic Launches Claude Opus 4.5 With Major Coding and Reasoning Upgrades

Anthropic has rolled out Claude Opus 4.5, its latest flagship AI model that claims to dethrone competitors in coding and complex reasoning. Available today, the update brings significant gains to software engineering tasks and computer control, arriving just as the rivalry with OpenAI heats up.



The company says Opus 4.5 sets a new high-water mark on the SWE-bench Verified test for real-world coding. Early testing suggests the model is far better at debugging multi-system issues and handling ambiguity without needing constant hand-holding. This release puts Anthropic in direct competition with the recent GPT-5.1 update, which also focused heavily on reasoning improvements.


A major addition for developers is a new "effort parameter" in the API. This allows users to choose between speed or deep thought. At its highest setting, Opus 4.5 reportedly beats the faster Sonnet 4.5 model while using nearly half the tokens, making it a more efficient choice for heavy-duty data analysis and coding workflows.

You can access the new model immediately through the Claude apps for iPhone and Mac, as well as the web interface. Anthropic has also updated its desktop app to handle parallel sessions in Claude Code, letting users run local and remote tasks at the same time. That move ramps up pressure on other Mac-focused AI tools, including OpenAI's new ChatGPT Atlas browser.

The update also brings better resistance to prompt injection attacks. API pricing is set at $5 per million input tokens and $25 per million output tokens.

Please download the iClarified app or follow iClarified on Twitter, Facebook, YouTube, and RSS for more AI updates.


Read More