Anthropic officially released “Claude Opus 4.5,” their latest flagship model, on November 24, 2025. This completes the 4.5 series, following Sonnet 4.5 in September and Haiku 4.5 in October.
Reclaiming the Coding Crown
Opus 4.5 is positioned as “the best model in the world for coding, agents, and computer use,” released in response to recent competitive models like OpenAI’s GPT-5.1-Codex-Max and Google’s Gemini 3.
Key benchmark results:
- SWE-bench Verified: 77.2% (standard configuration), 82.0% (high-compute setting)
- OSWorld: 61.4% (real computer-use tasks)
- Surpassed internal coding benchmarks while cutting token usage in half
Technical Specifications
- Context Length: 200,000 tokens (same as Sonnet 4.5)
- Output Limit: 64,000 tokens
- Knowledge Cutoff: March 2025 (Sonnet 4.5: January, Haiku 4.5: February)
- Pricing: $5 per million input tokens, $25 per million output tokens (one-third the cost of previous Opus-class models)
Major Platform Deployments
Opus 4.5 became available on multiple major platforms at launch:
Microsoft Ecosystem
- Microsoft Foundry: Public preview in East US2 and Sweden Central regions
- GitHub Copilot: Available to Pro, Pro+, Business, and Enterprise users (promotional 1x premium request multiplier through December 5, 2025)
- VS Code (Agent, Plan, Ask, and Edit mode)
- github.com
- GitHub Mobile (iOS/Android)
- Copilot CLI
- Microsoft Copilot Studio: Integrated support
Other Platforms
- Databricks: Available in DBSQL and Agent Bricks
- Anthropic Services: Default model for Pro, Max, and Enterprise plans
New Features and Product Integrations
Alongside Opus 4.5’s release, Anthropic announced several new features:
Claude for Chrome
Expanded from pilot to general availability for all Max users, providing AI assistance within the browser.
Claude for Excel
Available to Max, Team, and Enterprise users, featuring specialized capabilities for spreadsheet operations.
Claude Code Upgrades
- Enhanced Plan Mode: Creates more precise problem-solving plans and adheres to them more directly
- Desktop App Support: Claude Code now supported in the desktop application
AI Safety Commitment
According to Anthropic’s system card, Opus 4.5 is “the most robustly aligned model we have released to date and, we suspect, the best-aligned frontier model by any developer.”
The evaluation measures a wide range of “concerning behavior” scores, including both cooperation with human misuse and undesirable actions that the model takes at its own initiative, continuing the trend towards safer and more secure models.
Industry Impact
The release of Opus 4.5, just weeks after Opus 4.1, demonstrates the rapid pace of innovation in generative AI.
The model is designed for professional software developers and knowledge workers like financial analysts, consultants, and accountants, emphasizing its capability to handle complex enterprise tasks.
Summary
Claude Opus 4.5 emerges as a model that significantly improves the balance of performance, pricing, and practicality. Key highlights:
- Performance: World-class in coding and agent capabilities
- Cost: One-third the price of previous generation
- Practicality: Easy integration into daily workflows via Chrome and Excel
- Ecosystem: Immediately available on major platforms
As AI competition intensifies, Anthropic demonstrates competitiveness in both pricing and performance, accelerating its full-scale entry into the enterprise market.