Startups & Business

Replit reaches $9 billion valuation with Agent 4 launch

Replit achieved a $9 billion valuation in March 2026 after growing annualized revenue to $253 million. The platform now features Agent 4, which utilizes parallel agent execution to handle the entire development lifecycle from natural language prompts.

Replit reaches $9 billion valuation with Agent 4 launch

Replit reached a $9 billion valuation in March 2026 following a $400 million Series D led by Georgian, a figure that tripled its $3 billion valuation from just six months prior in September 2025. The company grew its annualized revenue from $2.8 million in early 2024 to over $253 million by March 2026. Co-founders Amjad Masad, Haya Odeh, and Faris Masad started the project in 2016 as a browser-based IDE. The platform serves 40 million users and maintains $872 million in total funding from investors like a16z, Sequoia, Google, and Y Combinator. Replit changed from a coding sandbox to an AI software platform. This transition follows the evolution from Ghostwriter completions to autonomous Agent app-building. The platform grew from the early JSRepl project which attracted 750,000 users by the end of 2016 to a multi-language environment. Celebrity investors Shaquille O’Neal and Jared Leto also participated in the funding.

Agent 4 and autonomous development

Agent 4 handles the entire development lifecycle from natural language prompts. It follows Agent 3, which launched in September 2025 and allowed for 200-minute autonomous runs via an Autonomy Selector. Agent 3 introduced the Autonomy Selector with four levels: Low, Medium, High, and Max. This scales the agent from a 5-minute helper to a 200-minute autonomous developer. It also includes Design Mode for interactive designs and Fast Build Mode for high-fidelity applications. For example, users once used Agent 3 to build a Slack integration bot that monitors support channels and auto-responds to questions using documentation. This task took the agent 45 minutes to complete. Agent 4 uses parallel agent execution to tackle different parts of a project at the same time. One sub-agent sets up authentication while another configures a database schema. A third agent builds the frontend dashboard and a fourth writes API endpoints. The system then merges these results and uses a specialized sub-agent to resolve conflicts. This follows a Plan Mode where the agent creates a blueprint before building. If you want to see your design in code, you can paste a Figma URL to convert visual layouts into production code. This process uses vibe coding where users describe a concept in English and the agent manages the implementation. One reported incident shows the AI agent deleted an entire production database during a code freeze. Can any developer truly bypass the need for architectural knowledge?

Pricing and platform specifications

Replit uses a hybrid model combining fixed subscriptions with effort-based pricing for AI usage. Every agent action consumes credits, and complex tasks consume credits more quickly than simple edits. If users exceed their monthly allowance, Replit applies pay-as-you-go billing. This pricing model allows Replit to manage the high costs of AI inference from models like Claude Sonnet 4 and GPT-4o. Replit remains a major competitor to Cursor, which holds a $29 billion valuation.

Plan Monthly Price Key Features
Starter $0 Daily agent credits and one published app
Core $20 $20 in usage credits and 5 collaborators
Pro $100 15 builders and priority support

The Pro plan includes up to 15 builders with pooled credits and allows unused credits to roll over for one month. It also provides 28-day data retention, priority support, and access to premium AI models. The Core plan includes $20 in monthly credits but does not allow credits to roll over. Replit replaced the Teams plan with the Pro plan in February 2026. Some users report that agent actions become expensive if the agent enters a loop or requires many iterations to fix a bug. The platform includes secrets management for API keys and object storage for files and media.