Anthropic Unveils Opus 4.5 with Expanded Claude Tools and New Infinite Chat Feature

Key Points

  • Anthropic releases Opus 4.5, enhancing coding, computer use, and office‑task performance.
  • Claude for Chrome extension becomes available to all Max plan users.
  • New infinite chat feature removes context‑window limits for paying Claude users.
  • Claude for Excel is now generally available to Max, Team, and Enterprise customers, supporting pivot tables, charts, and file uploads.
  • Internal tests show a 20% accuracy increase and a 15% efficiency gain with Opus 4.5.
  • Anthropic labels Opus 4.5 its safest model, outperforming Gemini 3 Pro in prompt‑injection resistance.
  • The model is accessible via Anthropic’s apps and API, with pricing starting at $5 per million tokens.

Anthropic's Opus 4.5 model is here to conquer Microsoft Excel
A table comparing Opus 4.5's efforts in various benchmarks.

A table comparing Opus 4.5's efforts in various benchmarks.

Opus 4.5: A Significant Model Upgrade

Anthropic announced that its flagship language model, Opus, has been upgraded to version 4.5. The new model builds on Opus’s reputation for strong coding assistance, computer‑use tasks, and office‑productivity capabilities, delivering state‑of‑the‑art performance across these domains.

Broader Availability of Claude Tools

Alongside the model release, Anthropic is extending the reach of several Claude‑branded tools. The Claude for Chrome extension, previously limited, is now available to all users of Anthropic’s Max plan, allowing seamless AI assistance directly within the browser.

In addition, Claude for Excel—a sidebar integration that brings the chatbot into Microsoft Excel—has been made generally available to Max, Team, and Enterprise customers. The Excel integration supports pivot tables, chart creation, and file uploads, enabling users to leverage AI assistance for complex spreadsheet tasks without leaving the application.

Introducing Infinite Chat

Anthropic’s newest feature, called “infinite chat,” addresses a common limitation of large‑language‑model interactions: the context‑window cap. For subscribers who pay to use Claude, infinite chat prevents the conversation from being cut off, preserving continuity across files and separate chat sessions. Anthropic described this capability as one of the most requested enhancements from its user base.

Performance Gains and Safety Improvements

Early internal testing of Opus 4.5 showed a 20 percent improvement in accuracy and a 15 percent boost in efficiency on tasks evaluated by Anthropic’s teams. The model also advances agentic workflow capabilities, allowing it to refine its own processes more effectively.

Anthropic emphasizes that Opus 4.5 is its safest model yet. According to the company’s own evaluations, the model better rejects prompt‑injection style attacks, outperforming even Google’s Gemini 3 Pro in this safety metric.

Availability and Pricing

Opus 4.5 is now accessible through all of Anthropic’s applications and via the public API. Pricing for the new model begins at five dollars per million tokens, providing a clear cost structure for developers and enterprises looking to integrate the upgraded capabilities.

Source: engadget.com