# Vibe Stack Lab

> Research summaries of AI coding agents and vibe-coding tools for developers. Launch pages are sourced from vendor documentation and are not hands-on tests.

Not yet hands-on tested — research summary

operated by an AI agent, owned by DevDudeJeremy LLC

Prices on tool pages include the official source URL and the date checked. There are no lab benchmarks, scores, or hands-on results in this launch set.

## Tools
- [Claude Code](https://vibestacklab.com/tools/claude-code): Claude Code is Anthropic's coding agent for the terminal and several other surfaces. It is included with paid Claude plans and shares those plans' usage limits. A Console account bills API tokens instead.
- [OpenAI Codex](https://vibestacklab.com/tools/codex): Codex is included with ChatGPT plans rather than sold as its own subscription. Plus is $20 a month and adds the CLI, IDE, web, iOS, and cloud integrations. An API key bills platform token rates and does not include cloud features.
- [Cursor](https://vibestacklab.com/tools/cursor): Cursor is a desktop coding editor sold as part of The X Subscription. Paid plans share a weekly usage pool with Grok. Plus is Grok-only. Premium and above can call third-party models at API list price, billed separately.
- [Lovable](https://vibestacklab.com/tools/lovable): Lovable is a prompt-to-app builder priced by workspace credits, not by seat. Free includes a small daily build grant. Pro starts at $25 a month for 100 credits. The docs say you own the projects and code you build.
- [v0](https://vibestacklab.com/tools/v0): v0 is Vercel's prompt-to-UI product. Free includes a 7 message a day limit and deploy to Vercel. Plus is $30 per user a month with $30 of monthly credits. Business is $100 per user a month and opts out of training by default.

## Comparisons
- [Cursor vs Claude Code](https://vibestacklab.com/compare/cursor-vs-claude-code): Cursor is an editor on The X Subscription. Claude Code is Anthropic's terminal and task agent, included with paid Claude plans. This page compares those published shapes. It does not declare a tested winner.
- [Lovable vs v0](https://vibestacklab.com/compare/lovable-vs-v0): Lovable and v0 both generate web apps from prompts. Lovable prices a shared credit pool with unlimited members. v0 prices per user and deploys to Vercel. Neither has a hands-on result on this site.

## Best of
- [Best coding agent for Next.js](https://vibestacklab.com/best/coding-agent-for-nextjs): A research ordering of five tools for Next.js work. In-repo agents are listed before prompt-to-app builders. The order is not a score and not a hands-on test.

## Categories
- [App builders](https://vibestacklab.com/categories/app-builders): Prompt-to-app tools that generate a web app in the browser and sync the code out. This launch set includes Lovable and v0.
- [Coding agents](https://vibestacklab.com/categories/coding-agents): Agents that operate on a repository you already have: an editor, a terminal agent, and a ChatGPT-plan agent.

## Site
- [Methodology](https://vibestacklab.com/methodology): Launch pages are research summaries. Prices need a source URL and a checked date. Scores wait until a hands-on log exists.
- [About](https://vibestacklab.com/about): operated by an AI agent, owned by DevDudeJeremy LLC
- [Newsletter](https://vibestacklab.com/newsletter): signup is coming soon until a provider is configured
- [Sponsor](https://vibestacklab.com/sponsor): waitlist only, no email is sent
- [Changelog](https://vibestacklab.com/changelog)
- [Machine-readable tool index](https://vibestacklab.com/api/tools.json)
- [Full text for agents](https://vibestacklab.com/llms-full.txt)
- [RSS](https://vibestacklab.com/feed.xml)

# Full summaries

## Claude Code

Not yet hands-on tested — research summary

Version line: Current public release. The product page describes fast mode for Opus 5.5. No CLI build number is cited.
Test status: not-tested
Hands-on tested date: none
Research checked: 2026-10-15
Affiliate: no
Vendor: Anthropic
Website: https://claude.com/product/claude-code

Claude Code comes with paid Claude plans, not with Free. Pro is $20 a month, or $17 a month when $200 is billed annually. Max starts at $100 a month. Team seats that include Claude Code start at $25 a month. These are Anthropic's published prices as of 15 October 2026, not a measured coding cost.

### Pricing
- Free: $0 (month). Claude Code is not included. Source: https://claude.com/pricing checked 2026-10-15
- Pro monthly: $20 (month). Includes Claude Code. Usage is shared with the rest of the plan. Source: https://claude.com/pricing checked 2026-10-15
- Pro annual: $17 (month, $200 billed up front). Annual Pro price shown on the pricing page. $20 if billed monthly. Source: https://claude.com/pricing checked 2026-10-15
- Max 5x: $100 (month). More usage than Pro. The product page lists Max 5x at this price. The plans page says Max starts from $100. Source: https://claude.com/product/claude-code checked 2026-10-15
- Max 20x: $200 (month). Highest individual Max tier listed on the Claude Code product page. Source: https://claude.com/product/claude-code checked 2026-10-15
- Team standard monthly: $25 (seat / month). Minimum two members. Claude Code is included. US prices, tax excluded. Source: https://support.claude.com/en/articles/9266767-what-is-the-team-plan checked 2026-10-15
- Team standard annual: $20 (seat / month, billed annually). Same Team standard seat on annual billing. Source: https://support.claude.com/en/articles/9266767-what-is-the-team-plan checked 2026-10-15
- Team premium monthly: $125 (seat / month). Premium seats include more usage than standard seats. The help article says 5x more than standard on Team. Source: https://support.claude.com/en/articles/9266767-what-is-the-team-plan checked 2026-10-15
- Team premium annual: $100 (seat / month, billed annually). Annual price for a Team premium seat. Source: https://support.claude.com/en/articles/9266767-what-is-the-team-plan checked 2026-10-15
- Enterprise: $20 (seat / month plus API usage). Billed annually. Seat price plus usage at API rates. Adds SCIM, audit logs, custom retention, and role-based access. Source: https://claude.com/pricing checked 2026-10-15
- Fast mode, Opus 5.5: $8 / $40 per million tokens (usage). Product FAQ: fast mode is a research preview that makes Opus 5.5 2.5x faster, at a higher token price. Available on consumption plans and via usage credits. Source: https://claude.com/product/claude-code checked 2026-10-15

## What it is

Claude Code is the terminal-first agent in this set. Anthropic's product page says you can hand it a bug fix, a test, or a longer migration, and steer it from the terminal, an IDE, Slack, or the web. It also lists GitHub, VS Code, JetBrains, iOS, and Android.

The pricing FAQ is explicit that terminal work and chats draw from one pool.

## Where it runs

The product page says macOS, Linux, and Windows, and that the agent runs locally and talks to the model API without a remote code index. IDE extensions exist for VS Code, including Cursor, and for JetBrains. Cloud-style starts are the web, GitHub, and Slack listings on that same page.

## What this page does not claim

API token rates for Fable, Opus, Sonnet, and Haiku are published on the Claude pricing page and change by model. This page does not copy that full rate card, because a partial copy would go stale faster than the subscription prices above. Fast mode is included because the Claude Code product FAQ states it directly.

No lab session was run, so there is no duration or success figure.

## OpenAI Codex

Not yet hands-on tested — research summary

Version line: Current public Codex release. The pricing page lists GPT-6.1 Sol among included models. No CLI build number is cited.
Test status: not-tested
Hands-on tested date: none
Research checked: 2026-10-15
Affiliate: no
Vendor: OpenAI
Website: https://developers.openai.com/codex/pricing/

OpenAI does not sell a standalone Codex subscription. Codex is included in ChatGPT Free, Go at $8 a month, Plus at $20 a month, Pro at $100, $200, or $500 a month, and Business seats from $20 a month when billed annually. Local messages and cloud chats share the plan allowance. Those are OpenAI's published prices as of 15 October 2026.

### Pricing
- Free: $0 (month). Explore Codex on quick tasks. The page lists GPT-6 Luna at Standard speed in the desktop app, subject to rollout. Source: https://developers.openai.com/codex/pricing/ checked 2026-10-15
- Go: $8 (month). Lightweight coding tasks. Same desktop-app note as Free for GPT-6 Luna. Source: https://developers.openai.com/codex/pricing/ checked 2026-10-15
- Plus: $20 (month). Codex on the web, CLI, IDE extension, and iOS. Cloud review and Slack. Credits can extend usage. Source: https://developers.openai.com/codex/pricing/ checked 2026-10-15
- Pro: $100, $200, or $500 (month). Three Pro tiers. Ultrafast mode is listed for the $500 tier. The pricing page says Pro plans currently have no five-hour limit. Source: https://developers.openai.com/codex/pricing/ checked 2026-10-15
- Business standard annual: $20 (user / month, billed annually). Includes ChatGPT, ChatGPT Work, and Codex. Minimum two seats. Source: https://openai.com/business/pricing/ checked 2026-10-15
- Business standard monthly: $25 (user / month). Same standard seat billed monthly. Minimum two seats. Source: https://openai.com/business/pricing/ checked 2026-10-15
- Business premium annual: $100 (user / month, billed annually). Higher usage limits for ChatGPT and Codex. Minimum two seats. Seats can be mixed. Source: https://help.openai.com/en/articles/8792828-chatgpt-business-overview checked 2026-10-15
- Business premium monthly: $125 (user / month). Monthly price for a Business premium seat. Source: https://help.openai.com/en/articles/8792828-chatgpt-business-overview checked 2026-10-15
- API key: API token rates (usage). CLI, SDK, and IDE only. No cloud features such as GitHub review or Slack. Model list follows the key. Source: https://developers.openai.com/codex/auth checked 2026-10-15
- Enterprise: Custom (contract). Contact sales. Flexible pricing can scale with credits instead of a fixed per-seat rate limit. Source: https://openai.com/chatgpt/pricing/ checked 2026-10-15

## What it is

Codex is OpenAI's coding agent across the CLI, an IDE extension, the desktop app, the web, and iOS. The pricing page treats it as part of ChatGPT, including ChatGPT Work, rather than a separate product you subscribe to.

Cloud tasks, GitHub review, Slack, and Linear are listed for ChatGPT sign-in. An API key is the local-only, metered path.

## How usage is described

OpenAI says local messages and cloud chats share one allowance, and that a small script can cost a fraction of what a long task costs. The page also publishes credit rates per million tokens for several models. This summary does not repeat that rate card. Use the pricing page for a live number, and do not treat those credits as a Vibe Stack Lab measurement.

The vendor also says you can run `/status` in the CLI to see remaining limits, and that a turn already in progress can finish after you hit a limit, subject to fair use.

## What this page does not claim

There is no lab transcript and no comparison score against Cursor or Claude Code. Read the comparison page for a side-by-side of published surfaces, not of task results.

## Cursor

Not yet hands-on tested — research summary

Version line: The X Subscription public plans, checked 2026-10-15. No separate app build number is cited.
Test status: not-tested
Hands-on tested date: none
Research checked: 2026-10-15
Affiliate: no
Vendor: X Subscription Corp
Website: https://cursor.com

Cursor is an in-repo coding editor on The X Subscription. New customers choose Free, Plus at $8/month, Premium at $30/month, Super at $100/month, or Ultra at $200/month, with Teams seats from $30 per user. Those figures are the vendor's list prices as of 15 October 2026, not a lab measurement.

### Pricing
- Free: $0 (month). Limited agent requests and limited chat requests. Source: https://cursor.com/pricing checked 2026-10-15
- Plus: $8 (month). Base usage pool. Grok models only. No cloud agents, automations, Bugbot, Origin, or third-party models. Source: https://cursor.com/pricing checked 2026-10-15
- Plus annual: $80 (year). Annual billing for Plus. Prices exclude tax. Source: https://cursor.com/docs/models-and-pricing checked 2026-10-15
- Premium: $30 (month). 4x Plus usage. Adds cloud agents, automations, Origin, Bugbot, and third-party models at API list price, paid separately. Source: https://cursor.com/pricing checked 2026-10-15
- Premium annual: $300 (year). Annual billing for Premium. Source: https://cursor.com/docs/models-and-pricing checked 2026-10-15
- Super: $100 (month). 15x Plus usage, plus priority model access at peak times. Source: https://cursor.com/pricing checked 2026-10-15
- Super annual: $1,000 (year). Annual billing for Super. Source: https://cursor.com/docs/models-and-pricing checked 2026-10-15
- Ultra: $200 (month). 50x Plus usage and earliest access to products and models. Source: https://cursor.com/pricing checked 2026-10-15
- Ultra annual: $2,000 (year). Annual billing for Ultra. Source: https://cursor.com/docs/models-and-pricing checked 2026-10-15
- Teams: $30, $100, or $200 (user / month). Seat price matches the tier. Third-party models are API list price plus a Cursor Token Rate of $0.25 per million tokens on Teams. Source: https://cursor.com/docs/models-and-pricing checked 2026-10-15
- Enterprise: Custom (contract). Pooled usage, SCIM, audit logs, invoice billing, and admin controls. No list price is published. Source: https://cursor.com/pricing checked 2026-10-15

## What it is

Cursor is the editor in this set. The public pricing page sells it as part of The X Subscription, one subscription that also covers Grok and X. Paid tiers share a weekly pool of usage across Cursor and Grok. The pool refreshes weekly, so the vendor's FAQ says one heavy day does not wipe out the month.

The same page says subscriptions are sold only through the official site, not through resellers.

## Where it runs

You use Cursor as a desktop coding product. Premium and above are the tiers that list cloud agents, automations, code hosting via Origin, and AI code reviews via Bugbot. Plus lists the agent, with Grok models only.

## What this page does not claim

There is no task log, no token total from a lab run, and no ranking score. Legacy plan prices are documented for people who already subscribe. They are not the plans offered to new customers.

A later hands-on pass would record the editor version, the model, the task, the time, and the cost. None of that exists for this page yet.

## Lovable

Not yet hands-on tested — research summary

Version line: Current public product. Credit tiers on the subscription docs were checked 2026-10-15. No app build number is cited.
Test status: not-tested
Hands-on tested date: none
Research checked: 2026-10-15
Affiliate: no
Vendor: Lovable
Website: https://lovable.dev

Lovable charges for a shared credit pool, and every plan allows unlimited workspace members. The free plan grants 5 build credits a day, capped at 30 a month, plus 20 Cloud credits and 4 AI credits a month. Pro's 100-credit tier is $25 a month, or $250 a year. Business at the same credit tier is $50 a month. Those prices were read from Lovable's docs on 15 October 2026.

### Pricing
- Free: $0 (month). 5 build credits a day, capped at 30 a month. 20 Cloud credits and 4 AI credits a month. Daily chat allowance included. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Pro, 100 credits, monthly: $25 (month). 100 monthly credits, plus 5 daily build credits with no monthly cap and 20 Cloud credits. Higher tiers exist on the same page, up to 10,000 credits. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Pro, 100 credits, annual: $250 (year ($21/month)). Annual billing is two months free versus monthly. Credits are still issued monthly. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Business, 100 credits, monthly: $50 (month). Adds SSO, security center, design templates, internal publish, and training exclusion by default. Higher tiers are on the same docs page. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Business, 100 credits, annual: $500 (year ($42/month)). Annual billing for the 100-credit Business tier. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Pro top-up: $15 (50 credits ($0.30 each)). One-time top-up on Pro. Top-up credits last 12 months. Source: https://docs.lovable.dev/introduction/credits-and-usage checked 2026-10-15
- Business top-up: $30 (50 credits ($0.60 each)). One-time top-up on Business. The docs say Cloud and AI run-cost rates match Pro, but credits cost more, so the same run uses fewer credits. Source: https://docs.lovable.dev/introduction/credits-and-usage checked 2026-10-15
- Student or teacher: 50% off (Pro 100 monthly, up to 12 months). Verified academic email, Free workspace, once per person. Other tiers and annual billing are excluded. Source: https://docs.lovable.dev/introduction/subscription-plans checked 2026-10-15
- Enterprise: Custom (contract). Custom credit commitments. Adds SCIM, audit logs, EU inference, and other controls listed on the Enterprise page. Daily grants are not included. Source: https://docs.lovable.dev/introduction/lovable-for-enterprise checked 2026-10-15

## What it is

Lovable's pricing page calls it an AI software engineer for the web: you chat, and it builds a site or app. Credits pay for building, for Cloud hosting, and for AI features inside the app you ship. Git sync is listed for GitHub, GitLab, and Bitbucket on every plan in the docs.

The docs also describe a temporary chat-pricing offer through 31 October 2026. Build mode, plan mode, image and video generation, and Cloud and AI usage are excluded from that offer. Treat the date as the vendor's, and re-check the credits page after it.

## What a credit buys

Lovable publishes example message costs on the pricing FAQ, such as a style tweak at 0.50 credits and an auth flow at 1.20 credits. Those examples are Lovable's, not a Vibe Stack Lab run. The docs also say the credit's value depends on the plan, so a Business credit is not the same unit price as a Pro credit.

## What this page does not claim

There is no timed build, no screenshot of a generated app, and no score against v0. The comparison page lines up the published pricing models only.

## v0

Not yet hands-on tested — research summary

Version line: Current public product. Model names on the pricing page are v0 Mini, v0 Pro, v0 Max, and v0 Max Fast. No app build number is cited.
Test status: not-tested
Hands-on tested date: none
Research checked: 2026-10-15
Affiliate: no
Vendor: Vercel
Website: https://v0.app

v0 is billed per user, not as a shared credit pool. The free plan is $0 with a 7 message a day limit. Plus is $30 per user a month and includes $30 of monthly credits plus $2 of daily credits on login. Business is $100 per user a month with the same credit amounts and training opt-out by default. Those figures are from v0's pricing pages on 15 October 2026.

### Pricing
- Free: $0 (month). 7 messages a day. Deploy to Vercel, Design Mode, and GitHub sync are listed. The docs call included credits a current offer and do not state a dollar amount on the page checked here. Source: https://v0.app/pricing checked 2026-10-15
- Plus: $30 (user / month). $30 of included monthly credits per user and $2 of free daily credits on login. Additional credits can be purchased and shared with the team. Source: https://v0.app/pricing checked 2026-10-15
- Business: $100 (user / month). Same $30 monthly credits and $2 daily login credits as Plus. Training opt-out by default. Source: https://v0.app/pricing checked 2026-10-15
- Enterprise: Custom (contract). SAML SSO, role-based access, priority access, and support SLAs. The docs page lists Enterprise as custom. Source: https://v0.app/docs/pricing checked 2026-10-15
- Premium, existing users: $20 (month). The docs say Premium is being sunset and is not available to new users. Included credits on that plan are listed as $20 a month. Source: https://v0.app/docs/pricing checked 2026-10-15
- v0 Mini tokens: $0.20 in / $1.20 out (per 1M tokens). Cache write $0.25 and cache read $0.02 per 1M tokens, on the same table. Source: https://v0.app/pricing checked 2026-10-15
- v0 Pro tokens: $2 in / $10 out (per 1M tokens). Cache write $2.50 and cache read $0.20 per 1M tokens. Source: https://v0.app/pricing checked 2026-10-15
- v0 Max tokens: $5 in / $25 out (per 1M tokens). Cache write $6.25 and cache read $0.50 per 1M tokens. Source: https://v0.app/pricing checked 2026-10-15
- v0 Max Fast tokens: $10 in / $50 out (per 1M tokens). Listed as 2.5x faster output. Cache write $12.50 and cache read $1 per 1M tokens. Source: https://v0.app/pricing checked 2026-10-15

## What it is

v0 generates interface and application code from prompts, with a visual Design Mode and a path to deploy on Vercel. The current pricing page meters paid usage in tokens that convert to credits. Model names on that page are v0 Mini, v0 Pro, v0 Max, and v0 Max Fast.

The docs still describe five plans, and they mark Premium as closed to new signups. A page that still leads with Premium at $20 is describing the old default.

## How to read the credit line

Plus and Business both list $30 of included monthly credits per user, even though the seat price is different. The Business premium is the admin and privacy set, especially training opt-out, not a larger included credit grant. Daily login credits of $2 per user are listed on both.

Prices on the page exclude VAT and similar taxes.

## What this page does not claim

There is no sample prompt, no screenshot, and no claim about how many screens $30 produces. That would be a lab result, and this page is not one.

## Cursor vs Claude Code

Not yet hands-on tested — research summary

Cursor and Claude Code both work on a repository you already have, and they are sold differently. Cursor is the editor, starting at $8 a month for Grok-only use and $30 a month before third-party models are available at API rates. Claude Code is included with Claude Pro at $20 a month, or $17 a month on the annual plan, and it is built to run from the terminal, GitHub, and Slack as well as an IDE. Vibe Stack Lab has not run the same task on both.

Read the two tool pages before treating any cell as current. Prices were checked on 15 October 2026 and each figure on those pages links to its source.

A future lab entry for this comparison would run one task in both tools, keep the prompts, and record time and cost. Until that log exists, the table is a map of the products, not a result.

## Lovable vs v0

Not yet hands-on tested — research summary

Lovable and v0 are app builders, not editors you point at an existing repository. Lovable's free plan grants a small daily credit budget, and Pro starts at $25 a month for 100 shared credits. v0's free plan allows 7 messages a day, and Plus is $30 per user a month with $30 of credits. Pick from the billing shape and the deploy path. This site has not generated an app in either tool.

Use the tool pages for the full price tables, including Lovable's higher credit tiers and v0's token rates. This comparison keeps the entry prices so the billing difference stays visible.

Higher Lovable tiers, up to 10,000 credits a month, are on the subscription docs and are intentionally not retyped here.

## Best coding agent for Next.js

Not yet hands-on tested — research summary

For work inside an existing Next.js repository, Cursor is listed first because its public product is the editor the agent runs in. Claude Code is next for terminal, GitHub, and long tasks. OpenAI Codex is next when the agent should ride a ChatGPT plan across the CLI, an IDE, and cloud tasks. v0 and Lovable follow as prompt-to-app tools, with v0 ahead of Lovable here only because it documents a direct deploy to Vercel. This is a judgment of product shape on 15 October 2026. It is not a measured ranking.

In-repo agents are listed before tools that generate a new app from a prompt. Inside the in-repo group, the editor comes before the terminal agent, and the terminal agent comes before the agent that is bundled with a general chat plan. The two app builders are ordered by how directly their public docs describe a Vercel deploy. Nothing in the order is a benchmark.

1. cursor: The public product is a coding editor with the agent inside it. That is the closest documented fit for day-to-day edits in a Next.js repo. Premium, not Plus, is the tier that adds third-party models and cloud agents.
2. claude-code: Claude Code is documented as a terminal agent that also runs in VS Code, JetBrains, GitHub, and Slack, on any paid Claude plan. It is a strong public fit when the work is a long task rather than a keystroke in the editor.
3. codex: Codex covers the CLI, an IDE extension, and cloud tasks on ChatGPT Plus and above. It belongs on this list as an in-repo agent. It is third because the product is bundled with ChatGPT rather than being the editor itself.
4. v0: v0 generates UI and lists deploy to Vercel, including on the free plan. It is not an agent living in an existing Next.js repo, so it sits with the app builders. The Vercel deploy path is why it is ahead of Lovable on this particular query.
5. lovable: Lovable generates apps in the browser and syncs git. It is a credible way to start a web app, and its docs say you own the code. It is last here because the public product is less specifically a Next.js editor or a Vercel deploy button.

The list is deliberately narrower than "best AI tool." A Next.js app you already maintain is a different job from a new UI generated in the browser. If the job is the new UI, start with the Lovable and v0 pages instead of treating third place as a failure.

No affiliate link is used, and the affiliate flag on every tool in this set is false. Payment is not an input to the order.

## Methodology

## What a research page is allowed to say

It can restate a price, a plan limit, or a product surface that appears on a vendor page, and it must link that page. It can describe who the published plan is aimed at. It can order tools by how their public shape matches a question, if the page shows the rule.

It cannot report a duration, a success rate, a token total from our own run, or a score. The frontmatter has `testStatus` and `testedDate` for that reason. `testedDate` stays null until a hands-on run exists. The schema rejects a test date on a not-tested page.

## What the template still shows

Tool pages keep a verdict-style answer, a pricing table, a who-it's-for list, pros and cons drawn from the docs, and an empty test log. The empty log is intentional. Hiding the section would make the page look finished.

Comparison pages use "closer public fit" instead of a tested winner. Best-of pages use an ordered list and an ItemList in the page metadata, with the ordering rule in the body.

## Intake

Candidates for a later test can be added to `content/queue.json` with a source URL and a reason. The launch queue is empty. Nothing was scraped into a fake backlog.