{"slug": "claude-turns-into-super-app-adding-web-browser-for-cowork", "title": "Claude turns into Super App, adding web browser for Cowork", "summary": "Anthropic is developing a built-in web browser for Cowork sessions in Claude Desktop, according to TestingCatalog, which would allow users to browse the web while Claude maintains context, bridging a feature gap with OpenAI's Codex. TestingCatalog also spotted a user-facing comparison mode that would let users manually compare responses from different Claude models, such as Opus and Fable 5. Both features are unreleased and no launch date has been announced.", "body_md": "Anthropic continues to enhance Claude Desktop, with recent app developments suggesting two potential additions that could move the product toward becoming a comprehensive AI super app.\n\nTestingCatalog has identified work on a built-in web browser for Cowork sessions. While a similar feature exists within Claude Code, it is currently not easily accessible from Cowork. The forthcoming implementation aims to integrate browsing directly into Cowork sessions, allowing users to open web pages alongside an active task while Claude maintains the surrounding context.\n\nThis enhancement would help Anthropic bridge a feature gap with OpenAI’s Codex, where browsing, coding, and agent workflows are increasingly integrated within the same desktop environment. Anthropic seems to view this as a broader opportunity for Claude Desktop, not merely an isolated browser feature.\n\nThis move aligns with a broader trend towards AI desktop super apps. Microsoft is progressing in this direction, Google is developing a Gemini desktop app, Meta is working on its own desktop application, and xAI has already launched its Grok app. With most major AI companies advancing towards persistent desktop environments that combine agents, tools, and browsing, Anthropic faces clear competitive pressure to match these developments.\n\nSimultaneously, TestingCatalog has observed work on a dedicated comparison mode. This feature would allow users to manually switch between regular chats and a view designed to compare responses from different [Claude](https://www.testingcatalog.com/tag/claude/) models. For instance, users could compare Opus with Fable 5 to determine if the output difference justifies spending more credits.\n\nThis does not appear to be an internal A/B testing interface. Instead, the comparison mode is being developed as a user-facing option that can be activated when needed and deactivated for normal conversations.\n\nBoth features remain unreleased, and no launch date has been announced. They were identified through references appearing in recent Claude builds.", "url": "https://wpnews.pro/news/claude-turns-into-super-app-adding-web-browser-for-cowork", "canonical_source": "https://www.testingcatalog.com/claude-turns-into-super-app-adding-web-browser-for-cowork/", "published_at": "2026-08-16 14:15:26+00:00", "updated_at": "2026-08-16 14:42:12.729517+00:00", "lang": "en", "topics": ["ai-products", "ai-agents", "artificial-intelligence"], "entities": ["Anthropic", "Claude Desktop", "Cowork", "TestingCatalog", "OpenAI", "Codex", "Microsoft", "Google"], "alternates": {"html": "https://wpnews.pro/news/claude-turns-into-super-app-adding-web-browser-for-cowork", "markdown": "https://wpnews.pro/news/claude-turns-into-super-app-adding-web-browser-for-cowork.md", "text": "https://wpnews.pro/news/claude-turns-into-super-app-adding-web-browser-for-cowork.txt", "jsonld": "https://wpnews.pro/news/claude-turns-into-super-app-adding-web-browser-for-cowork.jsonld"}}