Our verdict · App builder
The most complete platform in the builder group: the agent, a real IDE, databases with point-in-time restore, and publishing all live in one workspace, and it handles more languages than the web-only builders. The weak point is the bill: paid agent work is effort-based, so the cost of a request isn't known until the agent has done it.
Pick it if
- Solo builders who want database, hosting and code in one account
- Projects that outgrow a web-only stack or need a real IDE
- Data apps and internal dashboards with connectors
Skip it if
- You need to know a month's cost in advance
- You want a specific host or cloud from day one
- You only need a front end
Replit Agent works inside Replit's cloud workspace, which is the point: the agent, an editor, databases with point-in-time restore, connectors and publishing all live in one account. It builds web and mobile apps, dashboards and AI tools, and its docs now cover slides, videos and documents in the same project. This review compares Replit's published plans and docs against our six criteria; we didn't benchmark it.
Scores
- Speed to a first version4/5
Plan mode then build, with infrastructure handled; a little slower to a polished UI than Lovable.
- Depth before the wall4/5
A full IDE, many languages, checkpoints and background tasks let a project grow well past a demo.
- Backend and data5/5
Database with restore windows, publishing, storage and connectors all in one workspace.
- Ownership and exit3/5
Code moves to GitHub; the infrastructure around it has to be rebuilt elsewhere.
- Pricing predictability2/5
Effort-based pricing with no per-checkpoint rates published; Replit asks before paid work, but you can't price a month in advance.
- Team and production readiness4/5
Up to 15 builders on Pro, budgets and alerts; SSO, SCIM and audit logs on Enterprise.
Overall 3.7/5, the plain average of six criteria. Scored by Mahmoud Halat from published specs and docs, checked Nov 1, 2026. How we score.
Pricing: Free Mode, credits and effort
| Plan | Price (USD) | What you get |
|---|---|---|
| Starter | $0 | Daily agent allowance, limited cloud credits, 1 published app |
| Core | $20/mo ($18/mo annual) | $20 of credits, Free Mode, 1 background task, up to 5 collaborators, unlimited published apps, 7-day database restore |
| Pro | $100/mo ($90/mo annual) | Tiered credits that roll over two months, up to 10 background tasks and 15 builders, 28-day database restore, premium support |
| Enterprise | Custom | SSO, SCIM, audit logs, single-tenant and private networking options |
- Billed in
- Monthly credits; paid agent work is effort-based (priced per checkpoint by complexity)
- Free tier
- Starter: a daily agent allowance, limited cloud credits, one published app (30 days)
Source: replit.com/pricing, Pricing and plans help, AI billing docs. Prices checked Nov 1, 2026, before tax. Replit's help page lists Core annual billing as $204 a year ($17 a month) while the pricing page shows $18 a month billed annually; we show the pricing page. Third-party models are deducted at the provider's API rate. Plan Mode can incur charges even when no code changes.
Replit's pricing has three layers, and the third is the one that surprises people.
- The plan. Core is $20 a month and Pro $100, each with matching credits. Pro's credits are tiered and roll over for two months. Credit packs are discounted at volume: $300 of credit costs $290, $1,000 costs $950.
- Free Mode. Everyday agent work is free within an allowance that resets every five hours, with weekly caps on Core and Pro.
- Effort-based paid work. Beyond Free Mode, or in the Power and Max modes, the agent bills by complexity. A finished request becomes a checkpoint, and the checkpoint carries the cost. Replit asks for confirmation before paid work starts, shows each checkpoint's cost when you hover the usage icon, and lets you set budgets and alerts. What it doesn't do is publish what a given kind of request costs in advance; its docs point to an interactive demo instead.
Two smaller details from the docs: Plan Mode can be billed even when it changes no code, because planning is reasoning on the paid path; and third-party models such as Claude, ChatGPT and Gemini are deducted from your credits at the provider's public API rate. Credits also pay for published apps, storage and databases, so a busy app draws on the same balance as the agent.
One inconsistency to flag: Replit's pricing page shows Core at $18 a month billed annually, while its help page lists $204 a year ($17 a month). We show the pricing page's figure.
What it builds well
Replit is the most complete of the builders. Plan mode produces an ordered task list for you to approve before anything changes, the agent tests its own work and creates checkpoints you can roll back to, and databases come with restore windows of 7 days on Core and 28 on Pro. Connectors for BigQuery, Linear, Slack and Notion let an app pull real data. Because it's a full IDE, you can switch from prompting to editing whenever that's quicker, and you aren't limited to one web framework.
For a solo builder who wants database, hosting and code in one account, nothing else in the group needs fewer other services.
Where it struggles
The bill is the weak point: you can't price a month of work before you do it. Budgets and alerts limit the damage, but teams that need a fixed monthly cost will find Lovable's credits or a flat coding-agent subscription easier to plan around. Replit's UI output is also a step behind Lovable's polish for a first version, and the background-task limits (1 on Core, 10 on Pro) cap how much parallel agent work you can run.
Lock-in and leaving
Replit's docs describe GitHub connectivity: import a repository into a Replit app, modify it and push code back. So the code is portable. What isn't portable is everything Replit set up around it: the hosting, the database, secrets and connectors. Moving off Replit means recreating that infrastructure on another host, which is normal work for a developer and a real project for someone who chose Replit to avoid it.
Key facts
- Made by
- Replit
- Builds
- Web and mobile apps, dashboards, AI tools, plus slides, videos and documents in the same project
- Backend
- Agent configures databases; point-in-time restore of 7 days on Core and 28 on Pro; publishing built in
- Code export
- Import, modify and push code between Replit and GitHub
- Lock-in
- Code moves through Git; the hosting, database and secrets setup is Replit's own
Alternatives
- LovableFaster polished UI and simpler credit pricing
- Claude CodeWhen you want the agent in your own repo and infrastructure
- BoltBrowser builder with published token allowances
Head-to-head: Replit vs Lovable
Questions people ask
How much does Replit Agent cost?
Starter is free with a daily agent allowance and one published app. Core is $20 a month ($18 a month billed annually on the pricing page) with $20 of credits, and Pro is $100 a month ($90 annually) with tiered credits that roll over for two months. Paid agent work is billed by effort on top of Free Mode. Checked November 1, 2026.
What is Replit's effort-based pricing?
Paid agent work is priced by how complex the request is rather than a flat rate per message. When the agent finishes a request it creates a checkpoint, and that checkpoint carries the charge. Replit asks you to confirm before a paid action starts and shows each checkpoint's cost afterwards, but it doesn't publish per-checkpoint prices in advance.
What is Replit Free Mode?
No-cost agent work within a plan's allowance. On Core and Pro the allowance resets every five hours and has a weekly limit; Starter has a daily cap. Work beyond it, or in the paid Power and Max modes, uses credits.
Can I export a Replit app to GitHub?
Yes. Replit's docs describe importing, modifying and pushing code between Replit and GitHub. The code moves; the hosting, database and secrets configuration around it are Replit's and have to be set up again elsewhere.