By ByteBot
Anthropic released Claude Opus 4.5 on November 24, 2025, completing the 4.5 model series with the first AI to score over 80% on SWE-Bench Verified. This flagship model introduces Chrome browser automation and Excel spreadsheet integration—not just better benchmarks, but practical productivity tools that challenge Google and Microsoft on their home turf.
First Model to Break 80% on SWE-Bench Verified
Opus 4.5 scored 80.9% on SWE-Bench Verified, becoming the first model to exceed 80% on this real-world software engineering benchmark. This beats OpenAI’s GPT-5.1 (released November 12) and Google’s Gemini 3 (released November 18) in head-to-head coding performance.
The dominance extends beyond SWE-Bench. Opus 4.5 achieved 89.4% on Aider Polyglot multilingual coding tests compared to Sonnet 4.5’s 78.8%, leads on 7 out of 8 programming languages, and posted 59.3% on Terminal-bench and 66.3% on OSWorld computer use tasks. More impressive: at medium effort levels, it matches Sonnet 4.5’s best performance while using 76% fewer output tokens.
Chrome Extension Automates Browser Workflows
Claude for Chrome, available in beta to all Max plan users, shifts AI from chat-based assistance to direct browser automation. The extension manages multi-tab workflows, executes scheduled tasks, and includes platform-specific knowledge for Slack, Gmail, Google Calendar, Google Docs, and GitHub.
Anthropic’s internal teams use it to manage calendars, draft email responses, handle expense reports, and test website features. The Chrome extension reduced prompt injection attack success rates from 23.6% to 11.2% through security interventions. It blocks access to financial services, adult content, and pirated sites by default, requiring explicit approval for high-risk actions like publishing, purchasing, or sharing personal data.
Users drag browser tabs into Claude’s tab group, enabling the AI to view and interact with all tabs simultaneously. The “Ask before acting” permission mode creates an execution plan for approval, then runs the entire workflow independently within approved boundaries. Scheduled tasks automate recurring browser operations on a timer.
Excel Integration Delivers Measurable Productivity Gains
Claude for Excel, available to Max, Team, and Enterprise users, adds an AI sidebar that analyzes financial models, creates pivot tables and charts, and handles file uploads. Early testing by Anthropic showed 20% accuracy improvement and 15% efficiency gains when working with spreadsheets.
The Excel integration addresses what Anthropic calls the “black box” problem—when billions of dollars ride on a financial model’s output, analysts need to understand how the AI arrived at the answer. Claude for Excel shows its reasoning at the cell level, providing transparency that rivals like Microsoft Copilot don’t offer. This matters in finance, where explainable AI is mandatory, not optional.
The sidebar excels at model analysis, assumption updates, error debugging, template population, formula explanations, and multi-tab navigation. Users access it via keyboard shortcut: Control+Option+C on Mac, Control+Alt+C on Windows.
Infinite Chat Eliminates Context Window Errors
The “Infinite Chat” feature, available to all paid subscribers, solves the top user complaint: context window limit errors. When conversations reach the token limit, the model automatically compresses or summarizes older messages without resetting or alerting the user. This enables multi-day or multi-week coding sessions where the AI remembers initial constraints set weeks earlier.
Dianne Na Penn, Anthropic’s head of product management for research, explained: “Knowing the right details to remember is really important in complement to just having a longer context window.” The system preserves logically important constraints even when raw conversation text has been pushed far into the past.
Opus 4.5 supports 200,000 tokens by default, with special modes allowing up to 1 million tokens. Infinite Chat extends this indefinitely through intelligent compression.
Strategic Positioning Against Tech Giants
Anthropic completes its 4.5 series—Sonnet 4.5 in September, Haiku 4.5 in October, Opus 4.5 in November—while directly challenging competitors on their home turf. The Chrome extension competes on Google’s browser. The Excel integration challenges Microsoft Copilot. The SWE-Bench performance beats both OpenAI and Google in coding benchmarks.
More importantly, Opus 4.5 isn’t just about better benchmark numbers. Chrome automation and Excel integration deliver practical productivity gains that developers and analysts can measure. Anthropic’s approach to context windows—automatic compression versus alerting users—represents a more elegant solution than competitors offer.
Key Takeaways
- Opus 4.5 is the first model to score over 80% on SWE-Bench Verified (80.9%), beating GPT-5.1 and Gemini 3 in coding performance while using 76% fewer tokens at similar accuracy
- Chrome extension automates multi-tab browser workflows with scheduled tasks, platform-specific knowledge, and improved security (prompt injection attacks reduced from 23.6% to 11.2%)
- Excel integration shows 20% accuracy improvement and 15% efficiency gains with transparent cell-level reasoning that addresses the “black box” problem in financial modeling
- Infinite Chat feature eliminates context window errors through automatic compression, enabling multi-week coding sessions without losing initial constraints
- Strategic positioning challenges Google (Chrome), Microsoft (Excel), and OpenAI (coding benchmarks) with practical automation tools, not just benchmark improvements
