A cloud-based development platform that combines an IDE, AI Agent, deployment, and hosting in one browser tab. Its Agent 3 can autonomously build full-stack web and mobile apps from natural language, test them in a real browser, and fix issues without intervention—running up to 200 minutes per session and capable of generating other agents and automations (Slack bots, Telegram bots, scheduled workflows). The platform supports 50+ languages, React Native/Expo for mobile, built-in databases and auth, and one-click deployment with autoscaling. Pricing was restructured in February 2026: Starter (free), Core ($20/mo, $20 credits, 5 collaborators), Pro ($100/mo, Turbo Mode, up to 15 builders, credits roll over 2 months, 28-day DB restore). Teams plan is retired; existing Teams users were auto-upgraded to Pro. Cloud deployment prices dropped in August 2026 (autoscale compute from $3.20 to $0.60 per million units). SOC 2 Type II compliant, with 30+ connectors (Stripe, Figma, Notion, Salesforce) and custom MCP support.

Pros & Cons:

Pros: Agent 3’s self-testing and self-fixing loop in a real browser is a meaningful step beyond code generation—it catches issues that static generation misses, and can run autonomously for 200 minutes; the platform spans the full lifecycle (build, test, deploy, monitor) rather than just generation, with Monitoring Agent and Security Agent handling production issues and vulnerability scanning; mobile development is genuinely supported (React Native/Expo, simulator previews in the workspace), not an afterthought; deployment costs are now competitive—autoscale compute at $0.60/million units and database storage at $0.35/GiB make running published apps affordable; 30+ first-party connectors and custom MCP support reduce integration friction, and Pro’s credits roll over for 2 months, which is more forgiving than Core’s non-rolling credits.

Cons: Credit consumption is the biggest cost variable—Agent usage, deployments, and storage all draw from the same pool, and paid modes (Power/Turbo) consume credits based on task complexity, making monthly spend hard to forecast without setting budgets; Core credits do not roll over, and the jump from Core ($20) to Pro ($100) is significant for individuals who outgrow Core’s allowance but don’t need Pro’s team features; Agent 3’s autonomy can produce unexpected results—running for 200 minutes unsupervised means errors can compound before you review, and the “generate other agents” capability adds surface area that may need oversight; the platform is browser-based, so offline work and local-only workflows are not supported; independent third-party review volume on the post-Agent-3 product is still limited, and some 2026 pricing pages may not reflect the August cloud price cuts.

Who is it for?:Developers and non-developers who want to build, test, and deploy full-stack web or mobile apps without local setup; teams that want an AI Agent with real browser testing and production monitoring, not just code generation; builders targeting mobile apps (React Native/Expo) who want simulator previews and deployment in one place; users who need 30+ integrations and MCP connectors without building auth or database infrastructure from scratch. Not suited for offline-first workflows, teams that need local IDE control, or budget-sensitive users who cannot absorb variable credit consumption.