Anthropic Deploys Claude Code Web Platform While DeepSeek Advances OCR Compression Technology

Anthropic Expands Claude Coding Capabilities to Web Platform

Artificial intelligence company Anthropic has launched a beta research preview of Claude Code on the web, according to recent reports. The new platform enables developers to automate various coding tasks including bug backlogs, routine fixes, and parallel development work through Anthropic-managed cloud infrastructure. Sources indicate this represents a significant expansion of the company’s developer tools following their recent $183 billion valuation and preliminary approval of a $1.5 billion copyright settlement.

Anthropic Expands Claude Coding Capabilities to Web Platform
Enhanced Mobile Access and Security Features
Enterprise Growth and Strategic Partnerships
DeepSeek’s OCR Compression Breakthrough
Performance Metrics and Future Applications
Industry Implications and Future Directions

According to the company’s announcement, developers can now connect their GitHub repositories to the platform, describe their requirements, and allow Claude to implement solutions. The company stated that all Claude Code tasks will run in an isolated sandbox environment with network and filesystem restrictions for security purposes. Additionally, GitHub interactions are reportedly handled through a secure proxy service that restricts Claude to accessing only repositories explicitly authorized by users.

Enhanced Mobile Access and Security Features

Alongside the web platform launch, Anthropic is making Claude Code available on iOS devices, though analysts suggest the experiences may be refined as users submit feedback. The company emphasized the security measures implemented for the coding platform, noting that the isolated environment and restricted access protocols are designed to protect developer code and intellectual property.

The report states that Claude Code for the web is currently available for Pro and Max tier users. This launch follows Anthropic’s recent introduction of Claude Sonnet 4.5, which the company claims excels at coding tasks, and an updated version of its most affordable model, Claude Haiku 4.5. Industry observers note these developments position Anthropic more directly against competitors in the AI coding assistance space.

Enterprise Growth and Strategic Partnerships

Recent analysis suggests Anthropic is experiencing substantial enterprise demand growth. Reports indicate the company is projected to reach a $9 billion annual revenue run rate by year-end, with expectations of nearly tripling its annualized revenue by 2026. This growth trajectory appears to be driven largely by increasing enterprise adoption of their AI solutions.

According to sources, Anthropic recently announced its largest enterprise partnership to date—an expanded alliance with professional services firm Deloitte. The partnership will reportedly make Claude available to Deloitte’s workforce of approximately 470,000 professionals and involve developing industry-specific solutions powered by the AI platform. This landmark deal signals Anthropic’s strategic focus on enterprise markets as competition in the AI sector intensifies.

DeepSeek’s OCR Compression Breakthrough

Meanwhile, Chinese AI company DeepSeek has developed a new Optical Character Recognition system designed to compress large image-based text documents for more efficient AI processing. According to the company’s feasibility study, DeepSeek-OCR explores compressing long contexts through optical 2D mapping, which could significantly enhance how AI models handle document-based information.

The technical report describes the OCR system as comprising two main components: the DeepEncoder, which serves as the engine core, and a decoder. For training and evaluation purposes, DeepSeek researchers reportedly utilized 30 million PDF pages across approximately 100 languages, supplemented with synthetic diagrams, chemical formulas, and geometric figures to test the system’s capabilities across diverse document types.

Performance Metrics and Future Applications

According to the study findings, the OCR system achieved a decoding precision of 97% when text tokens numbered less than ten times the vision tokens. However, when vision tokens exceeded text tokens by twenty times, accuracy levels dropped to approximately 60%. These results provide important benchmarks for understanding the system’s performance characteristics under different document composition scenarios.

DeepSeek researchers suggest this study shows promising potential for developing future vision language models and large language models. The OCR technology emerges shortly after the company launched what industry analysts consider its most significant product release since V3 and RI. DeepSeek’s latest experimental model, V3.2-Exp, is described by the company as an intermediate step toward their next-generation architecture, indicating ongoing innovation in their AI development pipeline.

Industry Implications and Future Directions

The simultaneous announcements from Anthropic and DeepSeek highlight the accelerating pace of innovation in the AI sector. While Anthropic focuses on expanding practical coding applications through web and mobile platforms, DeepSeek’s research addresses fundamental challenges in processing visual document information. Industry observers suggest both approaches represent important vectors in AI development—applied tools for immediate productivity gains and foundational research for future capabilities.

According to market analysts, these developments reflect the broader competitive dynamics in the global AI industry, where companies are pursuing differentiated strategies to establish leadership positions. As both companies continue to advance their respective technologies, the industry appears poised for further innovation across both practical applications and underlying research frontiers.