New Features and Coding Capabilities
Claude Opus 4 and Claude Sonnet 4 set new standards for coding, advanced reasoning, and AI agents. Opus 4 leads on SWE-bench with 72.5% and Terminal-bench with 43.2%. It handles complex, long-running tasks and agent workflows. Both models include extended thinking with tool use in beta. This allows Claude to alternate between reasoning and tool use to improve responses. New capabilities allow the models to use tools in parallel and follow instructions more precisely. When developers grant access to local files, the models demonstrate improved memory. They extract and save key facts to maintain continuity and build tacit knowledge over time. Claude Code is now generally available for developers. It supports background tasks via GitHub Actions and native integrations with VS Code and JetBrains. It displays edits directly in your files for direct pair programming. The extensible Claude Code SDK allows developers to build their own agents and applications. The API includes the code execution tool, MCP connector, Files API, and prompt caching for up to one hour. Claude Opus 4 excels at coding and complex problem-solving. It delivers sustained performance on long-running tasks that require focused effort and thousands of steps. This model allows for work that lasts several hours and expands what AI agents can accomplish. Rakuten validated these capabilities with an open-source refactor that ran independently for seven hours. Cognition notes that Opus 4 handles critical actions that previous models missed. Both models are 65% less likely to use shortcuts or loopholes on agentic tasks than Sonnet 3.7. Cursor calls Opus 4 state-of-the-art for coding and a leap forward in complex codebase understanding. Replit reports improved precision and dramatic advancements for complex changes across multiple files. Block calls it the first model to boost code quality during editing and debugging in its agent, codename goose. iGent reports Sonnet 4 excels in autonomous multi-feature app development and reduces navigation errors from 20% to near zero. Augment Code reports higher success rates and more surgical code edits. GitHub says Sonnet 4 will power the new coding agent in GitHub Copilot. Manus highlights improvements in following complex instructions and aesthetic outputs. Sourcegraph says the model shows promise as a substantial leap in software development.
Claude Sonnet 4 delivers 72.7% on SWE-bench. It provides an optimal mix of capability and practicality. Amazon Bedrock expanded the context window for Claude Sonnet 4 to 1 million tokens in public preview. This 5x expansion from 200,000 tokens lets users load entire codebases or large document sets in a single request. The expanded context window for Claude Sonnet 4 in Amazon Bedrock allows users to load entire codebases, which include source files, tests, and documentation, so the model can understand project architecture and identify cross-file dependencies across the system. Users in US West (Oregon), US East (N. Virginia), and US East (Ohio) can access this expansion. Prompts over 200,000 tokens incur approximately twice the token price for input and 1.5 times for output. For organizations requiring identity automation or audit logs, the Enterprise plan provides SCIM provisioning and custom data retention controls. The Enterprise plan also includes a 500,000-token context window. If your organization requires the ability to automate user lifecycles through SCIM, the Team plan will not satisfy those needs.
| Claude Plan | Context Window | Monthly Cost per Seat |
|---|---|---|
| Free | Limited | $0 |
| Pro | 200,000 tokens | $20 (annual) / $25 (monthly) |
| Team | 200,000 tokens | $20 (Standard) / $100 (Premium) |
| Enterprise | 500,000 tokens | Custom |
Opus 4 costs $15 per million input tokens and $75 per million output tokens. Sonnet 4 costs $3 per million input tokens and $15 per million output tokens. For prompts exceeding 200,000 tokens, Sonnet 4 costs $6 per million input tokens and $22.50 per million output tokens.
Enterprise Deployment and Consulting
PwC announced a collaboration in February 2026 to embed Claude, including Opus 4.6 and Sonnet 4.6, into regulated environments in finance and healthcare. Deloitte rolled out Claude to 470,000 employees. Cognizant deployed it to 350,000 employees. Infosys integrated Claude into its Topaz AI platform. Most competitors still review governance frameworks and remain unsure where to place their agentic bets. Anthropic is backing a new enterprise services firm with Blackstone, Hellman and Friedman, and Goldman Sachs. This move follows OpenAI’s announcement of its Frontier alliance with BCG, McKinsey, Accenture, and Capgemini. Regulated sector deployment relies on safety and governance positioning. KPMG partnered with Anthropic for Claude for Life Sciences to assist with scientific research and clinical workflows. Slalom announced a formal partnership with Anthropic in November 2024 focused on ethical AI deployment on AWS. This resulted in a live case study with United Airlines, where Slalom used Amazon Bedrock and Claude to build AI-powered flight update customization. TCS is working with Anthropic and expects a formal partnership announcement soon. Will the emergence of these services firms bypass traditional systems integrators?
