Replit AI Review: What It Does, Pricing, and Alternatives

Draft v0.1 — 2026-05-23 KST. content_status = qa_passed. Generated from templates/tool-page-template.md and walked through Section A of qa/adsense-seo-quality-gate.md. Meta description (≤ 155 chars): Replit AI builds and runs apps from prompts in a browser — here is what it does, the Starter/Core/Pro/Enterprise pricing, and how it compares to GitHub Copilot.

Quick verdict

• Best for: learners, hobbyists, students, prototype-stage founders, and anyone who wants a "build it in the browser" loop where the AI agent, the editor, the runtime, and hosting all live in one tab.
• Not ideal for: professional engineering teams whose work lives in a local IDE on a private codebase, or organizations whose data policy forbids running production code on a third-party hosted environment.
• Pricing model: freemium. Starter is free with daily Agent credits. Replit Core is $20/month billed annually with $25 of monthly Agent credits and up to 2 parallel agents. Replit Pro is $95/month billed annually with $100 of monthly Agent credits, up to 10 parallel agents, and access to the most powerful models. Enterprise is custom. Verified on replit.com/pricing on 2026-05-23 KST.
• Free plan: yes — Starter is free, includes "Free daily Agent credits," and "Publish up to 1 project."
• Last verified: 2026-05-23 (replit.com/pricing page-body read)

Source-freshness note (2026-06-25)

If you are evaluating Replit AI as a browser/cloud build-run-deploy workflow — where the agent (Replit AI / Replit Agent), the editor, the runtime, and hosting all live in one tab — a 2026-06-25 KST recheck confirms Replit's own surfaces for verifying that decision yourself are reachable and stable. The AI/Agent product page (replit.com/ai, title "Replit AI – Turn natural language into apps and websites", H1 "Replit Agent: Make apps & sites with natural language prompts") and the pricing page (replit.com/pricing, title "Pricing - Replit", H1 "Pricing") both returned full page bodies in the same pass, still surfacing the durable "Agent" / "Build" / "deploy" framing, with the pricing page referencing Agent credits.

• Match the workflow before the specs. Replit AI/App/Agent keeps the agent, the editor, the runtime, and hosting in one browser tab — you describe an outcome, the agent builds it, and you run and publish it without leaving the page. Decide whether a browser-based build → run → deploy loop is how your work actually ships, versus a local IDE on your own machine, before weighing any headline number.
• Confirm runtime and deployment ownership. Because the runtime and hosting are part of Replit's hosted platform, confirm that running and publishing your code on a third-party browser environment fits your data, security, and compliance posture before adopting it on sensitive work.
• Keep a human review step before production changes. The agent proposes, runs, and can deploy code in the same loop that wrote it; treat its output as a draft your team reads, tests, and approves before anything ships to production — not a finished release. The vendor's positioning is not a guarantee about any specific run.
• Verify pricing and plans at the official source. Plan names, prices, Agent-credit amounts, parallel-agent caps, usage limits, and which models each tier includes change between Replit revisions. Read them on Replit's official pricing and product pages rather than trusting any figure — including the ones quoted lower on this page — once it is more than ~90 days old.

Source-backed freshness note: a reachability/title/H1/marker recheck of Replit's own AI/Agent and pricing pages on 2026-06-25 KST (evidence in data/replit-ai-source-gate-2026-06-25.json). No new price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, accuracy, or superiority claim is made here; volatile specifics are routed to the official site, and vendor positioning is reported as vendor evidence only.

Source-freshness note (2026-06-20)

If you are evaluating Replit AI as a browser/cloud coding and app-building workflow — where the agent, the editor, the runtime, and hosting all live in one tab — a 2026-06-20 KST recheck confirms Replit's own surfaces for verifying that decision yourself are reachable and stable. The AI/Agent product page (replit.com/ai, H1 "Replit Agent – Make apps & sites with natural language prompts", read HTTP 200) and the pricing page (replit.com/pricing, title "Pricing - Replit", H1 "Pricing", read HTTP 200) both loaded as full pages, still surfacing the durable "Agent" / "Build" / "deploy" framing, with the pricing page referencing Agent credits.

• Verify the current commercial terms at the source. Plan names, USD prices, Agent-credit amounts, parallel-agent caps, usage limits, and which models each tier includes change between Replit revisions. Read them on Replit's official pricing and product pages rather than trusting any figure — including the ones quoted lower on this page — once it is more than ~90 days old.
• Generated apps and code still need human review before shipping. The agent proposes, runs, and can deploy code in the same loop that wrote it; treat its output as a draft your team reads, tests, and approves before it ships — not a finished release. The vendor's positioning is not a guarantee about any specific run.
• Vendor evidence only. These are Replit's own pages, confirming the browser-based agent build-and-deploy surface is reachable with the "Agent" / "Build" / "deploy" framing — not an independent ranking or benchmark. Confirm current pricing, Agent-credit definitions, plan availability, and model availability on the official pages before relying on specifics.

Source-backed freshness note: a reachability/title/H1/marker recheck of Replit's own AI/Agent and pricing pages on 2026-06-20 KST. No new price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, or superiority claim is made here; volatile specifics are routed to the official site, and vendor positioning is reported as vendor evidence only.

Source-freshness note (2026-06-14)

If you are weighing Replit AI as a browser-based build-run-deploy workflow rather than a local coding assistant, note that Replit maintains a first-party AI product surface (replit.com/ai, titled "Replit AI – Turn natural language into apps and websites", read HTTP 200 on 2026-06-14 KST) and a companion pricing page (replit.com/pricing, titled "Pricing - Replit", read HTTP 200 on 2026-06-14 KST). Both pages still surface the durable "Agent" and "Build, deploy" framing, with the pricing page referencing Agent credits.

• Workflow fit. Replit AI keeps the agent, the editor, the runtime, and hosting in one browser tab — you describe an outcome, the agent builds it, and you run and publish it without leaving the page. Evaluate it against whether a browser-based build → run → deploy loop matches how your work actually ships, versus a local IDE on your own machine.
• Agent output still needs human review. The agent proposes and can run and deploy code; those changes still need a human to read, test, and approve before they ship — especially when the same loop that wrote the code also publishes it. Treat agent output as a draft, not a finished release.
• Hosting and runtime live with the vendor. Because the runtime and hosting are part of the same hosted platform, confirm that running and publishing your code on a third-party browser environment fits your data and compliance posture before adopting it on sensitive work.
• Vendor evidence only. These pages show Replit offers a first-party browser-based agent build-and-deploy surface with the "Agent" / "Build, deploy" / Agent-credits framing. That is the vendor's own positioning, not an independent ranking — confirm current pricing, Agent-credit definitions, plan availability, and model availability on Replit's official pricing and product pages before relying on specifics.

Source-backed freshness note drawn from Replit's own product and pricing pages. No benchmark, ranking, price, quota, credit, speed, model-availability, or superiority claim is made here; vendor positioning is reported as vendor evidence only, and all volatile specifics are routed to the official pages.

What is Replit AI?

Replit AI is the AI feature layer inside Replit, a browser-based development platform. Unlike a local AI coding assistant (which assumes you already have an editor, a runtime, and a deploy target on your machine), Replit puts all four in one browser tab: the AI agent that generates and modifies code, the editor you read it in, the runtime that runs it, and the hosting that publishes it.

Replit's positioning has shifted across product revisions — most notably toward "Agent" as the dominant interaction model, where you describe an outcome in natural language and Replit's agent builds, tests, and deploys an app. The pricing page on 2026-05-23 reflects that shift: agent credits, parallel agents, and model access are the primary differentiators between tiers, not seat count or editor features.

• Vendor: Replit
• Official homepage: https://replit.com/ai (AI feature surface), https://replit.com/ (product)
• Category: AI Coding Assistants

Main use cases

• Use case 1 — Generating apps from natural-language prompts: describing an app idea ("a small CRUD tool for tracking bookings", "a static site that displays a dashboard", "a Discord bot that does X"), and letting Replit's agent scaffold, edit, run, and deploy it without leaving the browser.
• Use case 2 — Browser-based prototyping and learning: building small projects on a Chromebook, school computer, or borrowed machine — anywhere installing a local toolchain would be inconvenient or impossible. The "everything in the browser" model is genuinely friction-free for beginners.
• Use case 3 — AI-assisted publishing and hosting: moving a generated prototype to a live URL with Replit's hosting on the same tier that built it, instead of stitching together a separate hosting provider.

Pricing and plans

The values below were read directly from replit.com/pricing on 2026-05-23 KST. Replit's plan structure has been revised multiple times (Hacker → Core, Starter, Agent credits), so reconfirm with the official page before quoting these numbers more than ~90 days from now.

• Starter — Free. Includes free daily Agent credits and the ability to publish up to 1 project. Also supports creating slides, videos, and animations.
• Replit Core — $20/month billed annually (the page describes this as a 20% discount vs the standard monthly rate; the standard monthly rate without annual commitment was not visible in the page section read). Includes $25 of monthly Agent credits and the ability to work in parallel with up to 2 agents.
• Replit Pro — $95/month billed annually (the page describes this as a 5% discount vs the standard monthly rate; the standard monthly rate without annual commitment was not visible in the page section read). Includes $100 of monthly Agent credits, up to 10 parallel agents, and "Access to the most powerful models."
• Enterprise — custom pricing. Inherits all Replit Pro capabilities ("Everything in Pro") with additional enterprise terms.

Source: live page-body read of https://replit.com/pricing on 2026-05-23 KST. Standard month-to-month rates without annual commitment, region-specific pricing, current promotional offers, the exact list of frontier models on the Pro tier, and the precise definition of "Agent credits" should be re-read directly when needed.

When evaluating Replit for a team or organization, also verify directly:

• Whether the work you would do on Replit can run safely on a hosted browser-based environment under your data policy.
• The current month-to-month price (without annual commitment) for Core and Pro, in your local currency.
• The Agent-credit definition and "parallel agents" cap as they apply to the work you actually want to do — credit accounting has changed across Replit revisions.
• Data-handling, code-snippet retention, and training-data opt-out policies per plan tier.

Pros

• The entire dev loop — code, run, deploy — lives in the browser. No local toolchain. This is uniquely valuable for learners and beginners.
• Real free tier with daily Agent credits, sufficient for evaluation and casual hobby work.
• Pricing escalates by AI usage (Agent credits, parallel agents, model access) rather than by seat count. For a single developer or a small team using AI heavily, this scales more naturally than per-seat seat-based products.
• "Build-and-publish in one tab" is a real ergonomic win when the alternative is wiring together an editor, a runtime, a hosting provider, and CI for a prototype that may never be released.

Cons and caveats

• A hosted browser-based platform is not the right fit for proprietary or compliance-sensitive codebases. Running production code on a third-party platform is a procurement decision; verify Replit's data-handling and code-handling policies before adopting on sensitive work.
• AI-generated code can be subtly wrong (mishandled edge cases, off-by-one, missed null checks, insecure defaults, hallucinated APIs). Treat agent output as a proposal that requires human review and testing — particularly when the agent also runs and deploys the code.
• Code-generation tools have outstanding legal questions around training-data sourcing, license inheritance, and code attribution. Consult counsel before relying on Replit's AI output for license-sensitive work.
• Replit's plan structure has changed several times, including how Agent credits are counted and what "parallel agents" means. Older comparison articles can be stale.
• The standard month-to-month price for Core ($20/month with annual commitment) and Pro ($95/month with annual commitment) was not visible in the section of the pricing page read on 2026-05-23 — only the annual-billing equivalent rates were shown. Verify the actual month-to-month rate before committing.
• Outputs are not professional advice. Do not treat Replit AI responses or generated code as a substitute for licensed legal, security, accounting, financial, or medical counsel.

Alternatives

• GitHub Copilot — better for professional engineering work in an existing local IDE on a private codebase already on GitHub, where the editor and version control are not negotiable.
• Cursor — better for developers who want an AI-first local editor with agentic multi-file edits, rather than a hosted browser-based platform.
• Claude (general assistant) — better if your top need is a chat assistant for design, drafting, and code review across many tasks, not a hosted IDE that also runs and deploys your code.

Buyer control and the review boundary

If you are evaluating Replit AI as a buyer rather than a casual user, the deciding question is less "can the agent build an app?" than "who stays in control of what it produces, and where does the review boundary sit?" These are the durable control questions to ask. <span id="replit-ai-review-boundary-2026-06-27"></span>

• Buyer control over the workspace. Because the agent, editor, runtime, and hosting share one browser tab, decide which workspaces and apps the agent may modify, run, and publish before you adopt it.
• App and workspace review ownership. Name who reviews each generated app or change before it ships — the agent proposes, but a human owns whether it is correct.
• Generated-code acceptance boundary. Accepting an agent's diff is the start of review, not the end; treat its output as a draft, as Cons and caveats describes.
• Human review before production changes. Keep approval a hard gate before anything deploys, since the same loop that writes the code can also publish it. Use the same questions on Cursor vs Replit AI or GitHub Copilot vs Replit AI, or the AI Coding Assistants category.

Evergreen decision framing only. No price, plan, quota, model-availability, benchmark, ranking, speed, accuracy, superiority, or legal claim is made here; verify current plan inclusions and data-handling on Replit's official pages, and confirm how the agent fits your workspace and review process against your own practice.

Where to compare Replit AI next

If Replit AI is on your shortlist, the next question is usually "against what, and for which job?" These side-by-side pages are organized by workflow fit, not by a winner — each one walks through where one tool's shape suits a particular task better than the other. Replit AI's job is building, running, and shipping code (with the agent, editor, runtime, and hosting in one browser tab), so the most direct comparisons are other coding tools; the cross-category links below are included for completeness but are not direct substitutes.

Same job — building, completing, and shipping code:

• Cursor vs Replit AI — when the deciding factor is whether you want an AI-first local code editor working on files on your machine, or a hosted browser platform that also runs and deploys what it builds.
• GitHub Copilot vs Replit AI — when you are weighing in-IDE completion and chat layered onto your existing local editor and GitHub repo against an all-in-one browser build-and-publish loop.
• Tabnine vs Replit AI — when the question is a completion assistant that emphasizes in-editor code suggestions versus a browser environment where the agent scaffolds and ships a whole project.

Cross-category — different primary job, included for orientation, not as substitutes:

• Claude vs Replit AI — a general reasoning-and-coding chat assistant versus a hosted platform that also runs and deploys code; compare here only if you are still deciding which category fits your task.
• Gemini vs Replit AI — a broad general-purpose assistant versus a browser-based build-and-ship environment; different categories, useful only for category-level orientation.
• Notion AI vs Replit AI — workspace writing-and-knowledge assistance is a separate job from building and running apps; this page is for readers triaging between the two categories.
• Zapier AI vs Replit AI — connecting apps and automating multi-step workflows is a different category from writing and shipping code; included only for orientation.
• Grammarly AI vs Replit AI — prose editing versus in-browser coding; another cross-category reference, not a like-for-like coding comparison.
• Replit AI vs Jasper — a templated marketing-content writing workflow versus a code build-and-ship platform; different primary jobs, included only for category-level orientation.

To browse the whole field rather than a single head-to-head, start from the AI Coding Assistants category. These links are decision paths, not rankings — no benchmark, price, quota, speed, accuracy, or model-availability claim is made here; the comparison pages route any such specifics to the official sources.

Who should not use Replit AI

• Engineering teams whose primary work is on a private codebase in a local IDE, where moving to a hosted browser-based platform would be a regression.
• Organizations whose data policy or compliance posture prohibits hosted execution of source code on a third-party platform.
• Users whose top requirement is in-IDE inline completion inside an editor that Replit is not — pay for GitHub Copilot or Cursor instead.

Author selection rubric

Choose Replit AI when at least two of these are true:

• Your dev work is browser-first by choice or by constraint (Chromebook, school computer, low-spec laptop).
• You value generating and shipping a prototype in one tab more than you value a deeply tuned local IDE.
• Your project's data sensitivity is compatible with running on a hosted third-party platform.

Avoid Replit AI when any of these are true:

• You work on a private codebase that must stay local or on your own infrastructure.
• Your editor is non-negotiable (Visual Studio, JetBrains, VS Code, Neovim, Xcode) and you want AI inside it.
• Your governance team requires AI-coding posture beyond what Replit's tier publishes.

Sources

• Official AI feature page: https://replit.com/ai — recorded as src-replit-ai-needs-verify in data/sources.json with access_status = ok (URL only).
• Official pricing page: https://replit.com/pricing — recorded as src-replit-pricing-2026-05-23 in data/sources.json with access_status = ok after a 2026-05-23 page-body read; this is the source for every plan, USD price, and Agent-credit allocation quoted on this page.
• Official AI + pricing reachability recheck: https://replit.com/ai and https://replit.com/pricing — recorded as src-replit-ai-homepage-pricing-2026-06-14 in data/sources.json with access_status = ok after a 2026-06-14 KST HTTP 200 read of both (titles "Replit AI – Turn natural language into apps and websites" and "Pricing - Replit"; durable markers Agent / Build, deploy / Agent credits). Source of the "Source-freshness note (2026-06-14)" section only; used as vendor evidence for the durable workflow-fit framing — no price, credit, quota, model-availability, benchmark, ranking, or superiority claim is drawn from it.
• Official AI/Agent + pricing reachability recheck: https://replit.com/ai and https://replit.com/pricing — recorded as src-replit-ai-official-2026-06-20 in data/sources.json with access_status = ok after a 2026-06-20 KST HTTP 200 read of both (AI page H1 "Replit Agent – Make apps & sites with natural language prompts"; pricing page title "Pricing - Replit" / H1 "Pricing"; durable markers Agent / Build / deploy / credits). Source of the "Source-freshness note (2026-06-20)" section only; used as vendor evidence for the durable browser/cloud build-run-deploy workflow framing — no price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, or superiority claim is drawn from it.
• Official AI/Agent + pricing reachability recheck: https://replit.com/ai and https://replit.com/pricing — recorded as src-replit-ai-home-pricing-2026-06-25 in data/sources.json with access_status = ok after a 2026-06-25 KST WebFetch of both that returned full page bodies (treated as reachable/HTTP 200; the tool does not surface the raw status code), with evidence saved at data/replit-ai-source-gate-2026-06-25.json/.md (AI page title "Replit AI – Turn natural language into apps and websites", H1 "Replit Agent: Make apps & sites with natural language prompts", markers Agent / Build / deploy; pricing page title "Pricing - Replit", H1 "Pricing", markers Agent / credits). Source of the "Source-freshness note (2026-06-25)" section only; used as vendor evidence for the durable browser/cloud build-run-deploy workflow framing — no price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, accuracy, or superiority claim is drawn from it.

Internal links (at least 3)

• Category page: /ai-coding/
• Alternative tool: /tools/github-copilot/
• Comparison pages: /compare/cursor-vs-replit-ai/, /compare/github-copilot-vs-replit-ai/, /compare/tabnine-vs-replit-ai/, /compare/claude-vs-replit-ai/, /compare/gemini-vs-replit-ai/, /compare/notion-ai-vs-replit-ai/, /compare/zapier-ai-vs-replit-ai/, /compare/grammarly-ai-vs-replit-ai/, /compare/replit-ai-vs-jasper/

Disclosure

• Affiliate links: none.
• Sponsored content: none. Replit has no relationship to this page.
• Generative AI assistance: this draft was assembled with the help of an AI assistant working from a 2026-05-23 live read of the official Replit pricing page; every plan, price, and Agent-credit claim is constrained to wording visible on that page on that date.

Trademark notice

Replit is a trademark of Replit, Inc. Use here is referential only and does not imply endorsement, partnership, or affiliation.

Update log

• 2026-06-27 (qualified-traffic/decision-content increment — evergreen, NOT a new revenue page, NOT a deploy): added a compact "Buyer control and the review boundary" section after the "## Alternatives" section and before "## Where to compare Replit AI next", carrying the durable marker replit-ai-review-boundary-2026-06-27 once (as a <span id> anchor). The section helps AI Stack DB readers evaluate Replit AI by control/review boundaries — buyer control over the workspace and which workspaces/apps the agent may modify, run, and publish; app/workspace review ownership; the generated-code acceptance boundary; and human review before production changes — and reuses only this page's own existing context (the agent/editor/runtime/hosting-in-one-tab framing and the Cons and caveats anchor already on the page). Links only existing local routes already in this content package: /compare/cursor-vs-replit-ai/, /compare/github-copilot-vs-replit-ai/, and the /ai-coding/ category. No source was fetched. No price, plan, quota, model-availability, benchmark, ranking, speed, accuracy, superiority, or legal claim was added; no Gumroad/UTM/affiliate/sponsored/coupon/checkout link added. data/tools.json, data/sources.json, and last_verified_at are unchanged; content_status stays qa_passed. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages). Local increment only — not a deploy.
• 2026-06-25 (source-freshness refresh — LIVE on aistackdb.com via Cloudflare Dashboard static-assets flow, NOT a new revenue page): added a compact "Source-freshness note (2026-06-25)" section near the top (above the retained "Source-freshness note (2026-06-20)"), framing Replit AI as a browser/cloud build-run-deploy workflow with durable evaluation questions — Replit AI/App/Agent workflow fit, runtime/deployment ownership, a human review step before production changes, and official pricing/plan verification. Evidence: a 2026-06-25 KST WebFetch recheck where Replit's official AI/Agent page https://replit.com/ai returned a full page body (treated as reachable/HTTP 200; the tool does not surface the raw status code) with title "Replit AI – Turn natural language into apps and websites", H1 "Replit Agent: Make apps & sites with natural language prompts", and durable body markers Agent / Build / deploy, and the pricing page https://replit.com/pricing returned a full page body with title "Pricing - Replit", H1 "Pricing", and markers Agent / credits — recorded as src-replit-ai-home-pricing-2026-06-25 (access_status = ok, evidence in data/replit-ai-source-gate-2026-06-25.json/.md) and added to the replit-ai sources_used/Sources. Vendor evidence only, not an independent ranking; no price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, accuracy, or superiority claim added; no Gumroad/UTM/affiliate/sponsored/coupon link added; data/tools.json and last_verified_at unchanged; content_status stays qa_passed. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages). Live deployment was completed through the visible Cloudflare Dashboard static-assets flow; exact deployment/version and ZIP evidence are kept in PM/run-log surfaces.
• 2026-06-20 (source-freshness refresh — local source, NOT a new revenue page, no deploy): added a compact "Source-freshness note (2026-06-20)" section near the top (above the retained "Source-freshness note (2026-06-14)"), framing Replit AI as a browser/cloud coding + app-building workflow and emphasizing official verification of current pricing/plan/Agent-credit/usage-limit/model availability plus human review and testing before shipping generated apps or code. Evidence: a 2026-06-20 KST recheck where Replit's official AI/Agent page https://replit.com/ai returned HTTP 200 (H1 "Replit Agent – Make apps & sites with natural language prompts"; durable body markers Agent / Replit AI / Build / deploy) and the pricing page https://replit.com/pricing returned HTTP 200 (title "Pricing - Replit", H1 "Pricing", markers Agent / credits), recorded as src-replit-ai-official-2026-06-20 (access_status = ok) and added to the replit-ai sources_used/Sources. Vendor evidence only, not an independent ranking; no price, Agent-credit, quota, plan, model-availability, benchmark, ranking, speed, or superiority claim added; no Gumroad/UTM/affiliate/sponsored/coupon link added; data/tools.json and last_verified_at unchanged; content_status stays qa_passed. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages).
• 2026-06-14 (source-freshness refresh — local source, NOT a new revenue page, no deploy): added a compact "Source-freshness note (2026-06-14)" section after "## Quick verdict" (before "## What is Replit AI?"), framing Replit AI as a browser-based build → run → deploy agent workflow from Replit's own AI and pricing pages (src-replit-ai-homepage-pricing-2026-06-14, https://replit.com/ai + https://replit.com/pricing, titles "Replit AI – Turn natural language into apps and websites" and "Pricing - Replit", durable markers Agent / Build, deploy / Agent credits, access_status = ok after a 2026-06-14 KST HTTP 200 read of both). Covers workflow fit (agent + editor + runtime + hosting in one browser tab), that agent output still needs human review and testing before it ships, that runtime/hosting live with the vendor (confirm data/compliance posture), and that pricing/credits/plan/model availability must be confirmed on official pages — vendor evidence only, not an independent ranking. Added one source to data/sources.json and the replit-ai sources_used/Sources. No pricing row changed; no benchmark, ranking, price, quota, credit, speed, model-availability, or superiority claim added; no Gumroad/UTM/affiliate/sponsored/coupon link added; data/tools.json and last_verified_at unchanged. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages).
• 2026-06-04 (internal-link/traffic refresh — live route now serves the section; no new revenue page, no web fetch): added a "Where to compare Replit AI next" section (placed after "## Alternatives", before "## Who should not use Replit AI") linking all nine qa_passed Replit AI comparison pages — three same-category coding (/compare/cursor-vs-replit-ai/, /compare/github-copilot-vs-replit-ai/, /compare/tabnine-vs-replit-ai/) and six explicitly cross-category/not-direct-substitute (/compare/claude-vs-replit-ai/, /compare/gemini-vs-replit-ai/, /compare/notion-ai-vs-replit-ai/, /compare/zapier-ai-vs-replit-ai/, /compare/grammarly-ai-vs-replit-ai/, /compare/replit-ai-vs-jasper/) — plus the /ai-coding/ category buying map, framed as decision paths not rankings. Expanded the "Internal links" comparison row to the same nine routes (it previously listed only /compare/claude-vs-github-copilot/). No source was fetched and no volatile claim added: no benchmark, ranking, price, quota, model-availability, speed, or accuracy fact is asserted, and no commercial, affiliate, sponsored, coupon, or tracking-parameter CTA was added to this page. data/sources.json, data/tools.json, and last_verified_at unchanged. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages).
• 2026-05-23 (draft and qa pass): first local draft created from templates/tool-page-template.md. Live page-body read of https://replit.com/pricing on 2026-05-23 KST added concrete Starter (Free) / Core ($20/mo annual) / Pro ($95/mo annual) / Enterprise (custom) plan names, USD prices, and Agent-credit allocations. New source entry added (src-replit-pricing-2026-05-23, access_status = ok). data/tools.json pricing_model, pricing_summary, has_free_plan = true, confidence_score, last_verified_at, and content_status refreshed. Section A1/A2 of qa/adsense-seo-quality-gate.md satisfied. content_status advanced to qa_passed. Standard month-to-month rates without annual commitment remain a noted soft blocker — only the annual-billing equivalents were visible in the 2026-05-23 fetch.