Anthropic has released Claude Opus 5.5, pricing it 40% below its predecessor at $4 per million input tokens, with the automation platform Activepieces already wiring the model into its workflow builder. For any small business running AI tools day to day, cheaper top-tier reasoning means more of the complex, multi-step work can be automated without the bill creeping up.

Anthropic has released Claude Opus 5.5 at 40% below the cost of its predecessor, bringing the price down to $4 per million input tokens. Opus 5.5 is Anthropic’s flagship model, built for long, multi-step tasks such as coding, document review and complex research, and it is available now through Anthropic’s own plans and through third-party platforms that connect to Claude via an API key.

The model sits at the top of a three-tier family. Claude Haiku 4.5 handles fast, simple jobs, Claude Fable 5.1 covers everyday reasoning, and Opus 5.5 is reserved for tasks where a mistake further down the chain would cause real problems, such as an automated workflow that queries a database, reshapes the data and pushes it live. This tiering lets a business match the cost of the AI call to how much a wrong answer would actually cost them, something UK firms are increasingly weighing up as Claude overtakes Copilot as the most in-demand AI skill on UK job listings.

Opus 5.5 can also read up to 200,000 tokens of text in one go, roughly a long contract or an entire policy document, without needing it broken into chunks first. For a small firm’s finance or legal admin, that means a single prompt can hold an entire supplier agreement or HR handbook and still track a clause on page one against a definition on page fifty.

The Activepieces side of the announcement matters for anyone already using no-code automation tools. The company has fixed a billing error that was undercharging some calls routed through OpenRouter to Opus, and every connector registered on its platform now doubles as a tool Claude can call directly, so a workflow built once does not need rebuilding to plug in a smarter model. That follows a pattern of automation platforms wiring Claude directly into business tools, similar to how Pleo recently connected Claude, ChatGPT and Copilot to its expense software.

On price, Anthropic’s consumer plans remain unchanged: the Pro plan is $20 a month with five times the usage of the free tier, and the Team plan is $30 per user a month with higher limits and admin controls. The $4 per million token rate applies to businesses calling the model directly through the API or through a connected platform, not to the monthly subscription plans most small firms would use day to day.

What to do: if your business already uses Claude through a subscription, nothing changes this week. If you or a developer you use is building automations through a platform such as Activepieces, check which model each workflow is set to, and move anything handling genuinely complex, high-stakes tasks to Opus 5.5 now that it costs less than before. For teams newer to AI prompting generally, Semrush’s list of prompts for Claude and ChatGPT users is a practical place to start before scaling up to agentic workflows.