Replit AI — Browser-Native Cloud IDE With Autonomous Agent 4, Prompt-to-Live-URL, Credit-Metered

Not shut down. Replit (founded 2016, Amjad Masad CEO) raised $400M Series D at $9B valuation Mar 2026, 50M+ users, ~$525M ARR run-rate Apr 2026, gross-margin positive, explicitly independent — not acquired, not winding down. Category ai-productivity-data; pipeline_stage none. Flagship is Replit Agent 4 (early 2026, 10× Agent 3 speed claim) doing prompt→plan→scaffold→DB→auth→deploy in one browser tab. Not Cursor (AI-first IDE, dev-centric), not Bolt.new (WebContainer, no persistent VM/DB), not v0 (React/Next purist), not Lovable (DB-baked app-first), not Codex/Claude Code (terminal/cloud agent, no hosted IDE). 2026 trap: effort-based credit burn unpredictable (Economy/Power/Turbo 1×/2×/6×), no default spend cap, Core $25 credits gone in days on heavy debugging, free Repls sleep.

1. One-line positioning

Replit AI = zero-setup browser cloud IDE (Nix containers, 50+ langs) + Replit Agent 4 (autonomous full-stack builder with plan mode, live-browser self-test, checkpoints, parallel agents) + built-in Postgres/KV/auth + one-click Autoscale/static deploy + real-time multiplayer + mobile app; credit-metered Agent work (model calls + compute + deploy traffic) pooled per plan. Fits ai-productivity-data; pipeline_stage none (prompt→agent plan→build→self-test→deploy→share). 63% of users are non-developers — deliberately the “anyone ships software” tool, not the senior-engineer IDE.

2. Core features

Layer	Capability	Notes
Agent 4 (2026)	NL prompt → clarifies → plans → scaffolds frontend/backend/DB/auth → runs code, reads errors, self-heals → live-browser click-test → deploys; checkpoints + rollback; up to 10 parallel agents on Pro	Agent 3 (Sep 2025) had 200-min autonomy + $0.20 self-test loop; Agent 4 claimed 10× faster
Cloud IDE	Browser editor (syntax, terminal, file tree, split panes), Nix reproducible env, no npm/Docker setup, fork-and-run in <10s	Performance lags VS Code on >10k file repos
Languages/stacks	50+ langs (Py/JS/TS/Go/Rust/Java/Ruby), templates for React/Next/Flask/FastAPI/Express, SQLite built-in, Neon/Supabase/PlanetScale/Convex connectors
Deploy	Static (free low-traffic) / Autoscale ($1 base + usage) / Reserved VM ($7+), custom domain, SSL, `.repl.co` subdomain free	No CI/CD branching, no traffic-split — hobby-grade
Multiplayer	Google-Docs-style cursors, code chat, bounties, up to 15 builders + 50 viewers on Pro	Smoother than VS Code Live Share per AISO
AI surface	Inline completions (Ghostwriter lineage), chat explain/refactor, Figma→React, AI tutoring highlights, mobile app (iOS/Android) natural-language build
Enterprise	SSO/SAML/SCIM, single-tenant VPC, static outbound IPs, data warehouse conn, Databricks/Lakebase integration, RBAC

3. Pricing (2026-08, USD, credit-metered)

Tier	Price	Credits / Agent
Starter	$0	Limited daily Agent credits, 1 published app, basic AI, public Repls sleep on idle
Core	$25/mo or $20/mo annual	$25/mo usage credits (no rollover), up to 5 collaborators, 2 parallel agents, private Repls, always-on deploy, remove badge
Pro (replaced Teams Feb 2026)	$100/mo or $95/mo annual	$100/mo credits (1-month rollover), up to 15 builders + 50 viewers, 10 parallel agents, Turbo mode, 28-day DB rollback, priority support
Enterprise	Custom	SSO/SCIM/single-tenant/VPC/static IP, custom seats

Credit mechanics: Agent draws from one pool covering model calls + compute minutes + outbound data + deploy overage. Modes: Economy (~⅓ Power cost) / Power (default) / Turbo (Pro+, ~2× faster, up to 6× credit cost). Overage = pay-as-you-go, no default spend cap unless user sets one.

pricing_type = freemium-credit-metered; starting_price = $0 (practical paid floor $20/mo annual Core); free_tier = "yes; limited daily Agent credits, 1 published app, public Repls only, sleeps on idle, no private Repls"; pricing_note = "replit-feb-2026-teams-retired-pro-replaces-it-95-100; credit-pool-covers-agent-compute-deploy-egress; economy-power-turbo-1x-2x-6x-burn; no-default-spend-cap-set-manually; core-25-credits-burns-in-days-heavy-debug; autoscale-1-base-static-low-free-reserved-vm-7; 1-month-credit-rollover-pro-only"

4. Access Type

access_type: web-ide-plus-agent-plus-deploy

access_display: replit.com (browser) · 📱 iOS/Android app · 🔌 GitHub/Slack/Notion/Stripe connectors · (no local IDE plugin — it is the environment)

5. Reviews (2026, split profile)

Source	Score	Signal
G2	4.5/5 (350+ reviews)	Speed-to-MVP, zero setup, Agent quality → praised
Capterra	4.6/5	Non-coders shipping first app
Trustpilot	~3.1–3.5/5 (1.5k reviews)	Billing anger: $50–90 surprise overages, credit vanish, slow refund support
Product Hunt	Positive	“Fastest from idea to URL”, but agent drifts on complex arch
Reddit/HN	Split	Lovers = PMs/designers/hobbyists; haters = engineers on credit burn + agent-introduced-bugs-while-fixing

Praise: lowest barrier in category (no local env), Agent 4 self-test loop catches its own errors, deploy-in-click, multiplayer best-in-class for teaching, 63% non-dev user base proves accessibility.

Gripes: credit burn unpredictable, no default cap, agent coherent only to medium complexity (breaks on 30-file refactors), cloud latency vs local, Replit lock-in (Nix env hard to export to Vercel/Railway cleanly), Trustpilot support complaints, free Repls sleep.

6. Best for / Not for

Best for	Not for
Non-technical founders / PMs / designers prototyping MVP in an afternoon	Senior engineers on large existing monorepo (Cursor/Claude Code better)
Educators / students (zero setup, classroom mgmt)	Teams needing predictable flat-cost AI (credit meter surprises)
Hackathon / weekend-app / waitlist-scrape / internal-tool builders	Production-scale apps with CI/CD, traffic-split, compliance (use Vercel+Railway+Cursor)
Solo devs wanting DB+auth+hosting in one tab	Tight-budget users who won’t watch the meter (Core $25 evaporates)
Vibe-coding agencies shipping 5–10 small apps/mo	Xcode/Swift native mobile (Expo preview only)

7. Competitors

Tool	Lane	Starts
Bolt.new	WebContainer prompt-to-app, no persistent VM/DB	Free / $20
Lovable	Full-stack app-first, Supabase-baked	Free / $20–25
Cursor	AI-first IDE (VS Code fork), dev-centric	$20 Pro / $200 Ultra
v0 by Vercel	React/Next/Tailwind/shadcn, Vercel deploy	$0 / $30 Plus
GitHub Copilot	Inline completion + light agent	$10 Ind
Claude Code / Codex	Terminal/cloud agent, no hosted IDE	$20+ / ChatGPT-bundled
Glitch (legacy)	Browser IDE, weaker agent now	—