Anthropic has released its newest flagship model, Claude Opus 4.5, calling it the company’s most capable system yet and positioning it as the “best model in the world” for coding, agents, and computer-based tasks.
The launch arrives at a time when competition between AI labs, such as Anthropic, OpenAI, Google, Meta, xAI and others has been intensifying, with each pushing out upgrades at a rapid pace.
According to Anthropic, Opus 4.5 shows major gains in software engineering tasks, outperforming other frontier models on SWE-Bench Verified, a benchmark that tests real-world coding problems. The model is available starting today across Anthropic’s apps, API, and major cloud platforms, with updated pricing of $5 US per million input tokens and $25 for output tokens.
Beyond coding, Anthropic says Opus 4.5 performs better on research tasks and handles documents like spreadsheets and slide decks more effectively than previous versions. The company also released updates to the Claude Developer Platform, Claude Code, Chrome integrations, desktop app, and Excel tools—all designed to take advantage of Opus 4.5’s stronger reasoning and longer-running agent capabilities.
Early testers inside Anthropic described Opus 4.5 as being better at handling ambiguous instructions, navigating tradeoffs, and solving multi-system bugs. The company claims that in one internal assessment used to test engineering candidates, Opus 4.5 scored higher (within a two-hour limit) than any human candidate who has previously taken the test.
Anthropic also says the new model shows improvements in safety, particularly around resisting prompt-injection attacks, and argues that Opus 4.5 is its most “aligned” system so far. The model’s ability to find workarounds in airline-style customer service scenarios raised questions within the company about how AI could reshape various professions, including engineering.
Below, check out the video of Claud Opus 4.5 solving a puzzle game:
The holy grail of AI right now is building a system that reasons like a human and can run a computer by itself, without turning into anything resembling Skynet (save us John Connor). It seems every week we’re hearing new and bigger advancements. Are you scared or excited for what’s to come?
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
