Best Vibe Coding Tools
VibeToolRank tracks 40 tools across app builders, editors, and coding agents in this vibe-coding guide; 27 have been tested. We list app builders first, then editors, then agents — not a single overall scoreboard. 13 not yet tested.
This is a vibe-coding guide, not a single cross-category scoreboard. We still use the recorded overall scores from our tests, but we group tools in the usual prompt-to-app path: AI app builders first, then AI code editors, then coding agents. The highest overall score on this site may belong to an editor or a CLI agent; that does not make it the first tool we show for vibe coding. Scores inside each section are not one shared benchmark, and untested tools are marked not yet tested. For a raw scoreboard of the same three categories, see Best AI Coding Tools.
A cross-category ranking of app builders, AI code editors, and coding agents. Only tested tools receive scores. Last updated: 2026-11-12 · Methodology · Markdown
AI App Builders
Ordered by recorded overall score inside this section of the vibe-coding path. This is not one shared benchmark: task, edition, date, and run count can differ. Full AI App Builders ranking.
| # | Tool | What it is | Score |
|---|---|---|---|
| 1 | Bolt.new | In-browser AI app builder by StackBlitz. | 8.7cloud free plan |
| 2 | v0 | Vercel's AI app and UI generator. | 8.3cloud free plan |
| 3 | Base44 | No-code AI app builder. | 8.2cloud free plan |
| 4 | Lovable | Prompt-to-app builder for full-stack web apps. | 8.0cloud free plan |
| 5 | Replit Agent | Cloud IDE with an AI agent that builds and deploys apps. | 7.6cloud free plan |
| 6 | Google AI Studio | Google's free Gemini playground whose Build mode turns prompts into web apps. | 7.5in browser, free |
Not yet tested: Anything.
AI Code Editors & IDEs
Ordered by recorded overall score inside this section of the vibe-coding path. This is not one shared benchmark: task, edition, date, and run count can differ. Full AI Code Editors & IDEs ranking.
| # | Tool | What it is | Score |
|---|---|---|---|
| 1 | Kiro | Spec-driven agentic IDE from AWS. | 9.7Free CLI / AWS Builder ID |
| 2 | TRAE | ByteDance AI coding stack: TRAE IDE (trae.ai) plus the open-source trae-agent CLI (bytedance/trae-agent) for headless software-engineering tasks with BYOK models. | 9.0OSS CLI |
| 3 | Cursor | AI-native code editor. | 8.4desktop app, free plan |
| 4 | Devin Desktop (formerly Windsurf) | AI IDE formerly called Windsurf; renamed Devin Desktop on June 2, 2026. | 8.1desktop app, free plan |
| 5 | GitHub Copilot | GitHub's AI pair programmer for editors and the CLI. | 8.0desktop app, free plan |
| 6 | Zed | High-performance code editor with a built-in Agent; free forever on Personal with your own API keys. | 6.8Desktop, BYOK free |
Not yet tested: T3 Code, ThinkRail (JetBrains).
Coding Agents (CLI & Cloud)
Ordered by recorded overall score inside this section of the vibe-coding path. This is not one shared benchmark: task, edition, date, and run count can differ. Full Coding Agents (CLI & Cloud) ranking.
| # | Tool | What it is | Score |
|---|---|---|---|
| T-1 | Aider | Open-source AI pair programming in the terminal. | 9.3CLI, free model |
| T-1 | Claude Code | Anthropic's agentic coding tool for the terminal, IDE, desktop, and browser. | 9.3Anthropic Sonnet 5.5 API |
| T-3 | Cline | Open-source autonomous coding agent for VS Code. | 9.2CLI, free model |
| T-3 | GitHub Copilot CLI | GitHub's Copilot coding agent in the terminal, included in Copilot Free. | 9.2Copilot Free CLI |
| T-3 | OpenCode | Open-source AI coding agent for the terminal. | 9.2CLI, free model |
| T-6 | Memdoor | Lightweight open-source coding and automation agent in one CLI binary for macOS, Linux and Windows. Run it locally or over SSH, headless from scripts, CI or cron, with multi-step YAML workflows, approval gates and a decision model that trims token use. Bring your own key or a local model. | 9.1CLI, BYOK |
| T-6 | OpenHands | Open-source platform for software engineering agents (MIT). Self-host with Agent Canvas, a headless CLI, an SDK, or use OpenHands Cloud. Bring your own LLM key; Cloud also offers at-cost hosted models and a limited free BYOK tier. | 9.1headless CLI, BYOK |
| T-6 | Zerg Client | Zerg Terminal Client (ZTC): a CLI and macOS desktop AI coding agent that plans changes, edits code and runs tests in your project. Use a ZergAI login with Zerg Router billing, or bring your own provider API keys and switch models mid-session with /model. Headless on macOS, Linux and Windows; native terminal and desktop app currently Apple Silicon macOS. | 9.1Headless CLI, BYOK |
| 9 | Amp | Agentic coding tool from Sourcegraph. | 9.0cloud free plan |
| 10 | Reason | Autonomous cloud coding agent for GitHub that opens pull requests; its new Reason Network lets your coding agents talk to each other. | 8.6cloud free plan |
| T-11 | Alysis Code | Open-source AI coding agent (Apache-2.0 engine) with desktop app (macOS Apple silicon + Windows) and CLI (macOS/Linux/Windows). Code mode builds/fixes/tests in your repo; Work mode handles files/browser/apps. Bring your own model (OpenAI, Anthropic, Gemini, OpenRouter, Ollama, etc.), use Alysis credits, or run local models. Sandboxed by default; CLI supports one-shot headless tasks. | 8.4CLI, BYOK |
| T-11 | Harness (Autonomous) | Open-source command center that runs Claude Code, Codex and other coding agents across all your machines side by side. | 8.4CLI, free model |
| 13 | Jules | Google's asynchronous coding agent. | 8.0cloud free tier |
| 14 | OpenAI Codex | OpenAI's coding agent for CLI, IDE and cloud. | 7.4CLI, free model |
| 15 | Gemini CLI | Open-source terminal agent from Google; consumer users moved to Antigravity CLI in June 2026. | 6.1CLI, free API tier |
Not yet tested: Agent Manager, Conducted, Crewplane, Devin, Mirror (Reflection), Parallax (Conifer), psi, Roomote (formerly Roo Code), Ski, Wonder.
How we pick winners
Overall score: 0.4×Outcome + 0.15×(Speed+Cost+DX+Reliability). Editions are labeled where recorded. Scores are never sold as paid placements. On this vibe-coding guide, section order is the prompt-to-app path, not overall score. Read the methodology.
Detailed reviews
1. Bolt.new — 8.7/10
In-browser AI app builder by StackBlitz. Tool details
Pricing: Official pricing page read 2026-11-12: Pro from $25 (the page advertises saving up to 28% with yearly billing; confirm the annual total on the page before paying); Teams $30 per member, billed monthly; Enterprise custom. Pro adds no daily token cap, starts at 10M tokens/month, unused-token rollover while paid (up to two months per the FAQ), custom domain, and badge removal. · Free tier: Yes: Free forever, no card — 300K tokens/day and 1M tokens/month; unused Free tokens do not roll over; public and private projects; website hosting with Bolt branding and up to 333k web requests; 10MB upload limit (bolt.new/pricing, 2026-11-12).
Our T1 todo prompt gave a working preview in about 68s. Add, complete, delete, filters and localStorage persistence all passed with zero fixes, and it added an unrequested 'Clear done' button. The build used about 52K of the 1M free monthly tokens. Single run, not 3. cloud free plan
- Outcome
- 9
- Speed
- 9
- Cost
- 9
- DX
- 8
- Reliability
- 8
- Working todo app preview in about 68 seconds
- Every T1 check passed with zero fixes
- Free one-click publish to a .bolt.host URL
- Added an unrequested 'Clear done' feature
- Free plan is token-capped at 1M per month and 300K per day
- Custom domain and badge removal need Pro at $25/mo
- Signup needs a Cloudflare check and email confirmation
Who it’s for: Non-coders and fast prototypers who want a working web app from a prompt, entirely in the browser. Bolt Free suits in-browser StackBlitz builds with published Free token caps and free .bolt.host publish. Lovable Free uses a different credit meter and feature set. Overall scores came from separate 2026-10-16 single runs, not one head-to-head session. Do not rank Free capacity by comparing tokens to credits.
2. v0 — 8.3/10
Vercel's AI app and UI generator. Tool details
Pricing: Plus $30/user/mo; Business $100/user/mo; Enterprise custom · Free tier: Yes: 7 messages/day; new users get one-time $5 credits
Our T1 todo prompt was done in about 94s, with a preview link at 25s. v0 tested its own work in its browser, found a missing lib/utils file and fixed it. Add, complete, delete, filters and reload persistence all passed with no manual fixes. The build cost $0.64 of the $5 free credit. Single run, not 3. cloud free plan
- Outcome
- 9
- Speed
- 9
- Cost
- 7
- DX
- 7
- Reliability
- 8
- Done in about 94 seconds, with a preview link after 25 seconds
- Tested its own work, found a missing file and fixed it without help
- Every T1 check passed with no manual fixes
- Publish button available on the free plan
- Signup requires phone SMS verification
- The $5 free credit covers only about 7 builds like ours at $0.64 each
- v0's own pages disagreed on whether the $5 credit renews monthly or is one-time
Who it’s for: Developers, especially in the Next.js and Vercel ecosystem, who want fast, self-checking app generation from a prompt.
3. Base44 — 8.2/10
No-code AI app builder. Tool details
Pricing: Official pricing page read 2026-11-12 (annual view): Starter $16/mo, Builder $40/mo, Pro $80/mo, Elite $160/mo, billed annually. The page FAQ says annual billing saves 20% (monthly price × 12, charged once a year), which implies monthly prices of $20/$50/$100/$200; those monthly figures were not shown as numbers on the annual view we read. Enterprise custom. Free $0. · Free tier: Yes: Free plan with 25 message credits/mo, up to 5 message credits/day within that monthly total (official docs, 2026-11-12), 100 integration credits/mo, and up to 5 apps. Only the personal signup workspace gets free message credits; other Free workspaces get integration credits only.
Our T1 todo prompt gave a working preview in about 3.4 minutes. Add, complete, delete, All/Active/Done filters and reload persistence all passed with no manual fixes. Its own verifier failed 3 times on a leftover placeholder page, then fixed itself. It also added sign-in and put the list behind login without being asked. The build cost 1 credit. Single run, not 3. cloud free plan
- Outcome
- 8
- Speed
- 8
- Cost
- 9
- DX
- 8
- Reliability
- 8
- Every T1 check passed with no manual fixes
- The build cost only 1 of 25 free monthly credits
- Free publish to a *.base44.app subdomain
- Signup needs only email and a 6-digit code, no phone
- Slower than Bolt: about 3.4 minutes to a working preview
- Added sign-in and put the list behind login without being asked
- Its verifier failed 3 times on a leftover placeholder page before fixing itself
- Custom domains need a paid plan
Who it’s for: Non-coders who want a full app with built-in login and data from a prompt, and don't mind a slower build.
4. Kiro — 9.7/10
Spec-driven agentic IDE from AWS. Tool details
Pricing: Free $0 (50 credits/mo); Pro $20/user/mo (1,000 credits); Pro+ $40 (2,000); Pro Max $100 (5,000); Power $200 (10,000); add-on credits $0.04 each · Free tier: Yes: perpetual Free with 50 credits/mo — Auto, Claude Sonnet 4.5/4.0, and open-weight models (DeepSeek 3.2, MiniMax M2.5/M2.1, GLM-5, Qwen3 Coder Next); flagship Claude Opus / Sonnet 5.5 / GPT-5.6 need a paid plan
Three Free CLI T3 runs on 2026-11-02: each fixed the seeded cart.js discount bug so node --test passed 4/4 in 17–18s, leaving cart.test.js untouched and self-running the tests. Used about 0.80 of 50 monthly Free credits total ($0 spent) after Builder ID device-flow login; no API key. Free CLI / AWS Builder ID
- Outcome
- 10
- Speed
- 10
- Cost
- 10
- DX
- 8
- Reliability
- 10
- 3/3 Free CLI T3 runs fixed the bug in 17–18s with 4/4 tests passing
- Agent edited cart.js and ran node --test itself
- Used only about 0.80 of 50 monthly Free credits for three runs ($0)
- Free Builder ID session supports headless chat without an API key
- Free is 50 credits/mo and excludes Claude Opus, Sonnet 5.5, and GPT-5.6
- Official API keys / documented headless KIRO_API_KEY path need Pro or higher
- First login still needs interactive browser or device flow
- Headless Free runs need --trust-all-tools and --trust-workspace
Who it’s for: Developers who want a spec-driven agentic IDE/CLI from AWS, with a usable free tier for small tasks — and paid tiers that put several flagship models on one credit pool.
5. TRAE — 9/10
ByteDance AI coding stack: TRAE IDE (trae.ai) plus the open-source trae-agent CLI (bytedance/trae-agent) for headless software-engineering tasks with BYOK models. Tool details
Pricing: Pro $20/mo ($200/yr); Pro+ $60/mo; Ultra $200/mo; Lite $8/mo in select countries only · Free tier: Yes: TRAE IDE has limited monthly Auto-mode usage (no card claimed on docs). trae-agent OSS is free to self-host with your own model key.
trae-agent OSS CLI fixed our seeded cart discount bug in 3 of 3 runs (19–42s each) with a free Gemini key, never touched the test file. Cost $0. Overall 9.0 (outcome 10, speed 8, cost 10, DX 5, reliability 10). DX lost points because stock trae-agent sends Gemini tool results with role=tool, which the current Gemini API rejects until you patch google_client.py; a yaml config file is also required. OSS CLI
- Outcome
- 10
- Speed
- 8
- Cost
- 10
- DX
- 5
- Reliability
- 10
- Fixed the seeded bug in 3/3 CLI runs with a free model
- Did not modify the test file in any run
- OSS self-host / BYOK path needs no TRAE IDE account
- Fast once the Gemini client patch is applied (19–42s)
- Stock google_client uses role=tool and fails against current Gemini API until patched
- Requires a trae_config.yaml; CLI flags alone are not enough without a config file
- We scored the OSS CLI only, not the TRAE IDE free Auto mode
Who it’s for: Developers who want a ByteDance-built coding agent; IDE users get limited free Auto mode, and CLI users can self-host trae-agent with a free Gemini or OpenRouter key.
6. Cursor — 8.4/10
AI-native code editor. Tool details
Pricing: Via The X Subscription (prices exclude tax; checked 2026-11-06): Free $0 (limited); Plus $8/mo (Cursor models only; no third-party models; no own API keys); Premium $30/mo (4× Plus usage; third-party models at API list price from prepaid credit); Super $100/mo (15×); Ultra $200/mo (50×); Teams $30/$100/$200 per seat/mo. Annual: Plus $80/yr, Premium $300/yr, Super $1,000/yr, Ultra $2,000/yr. Legacy Pro $20/mo for existing subscribers only. Grok API on paid plans is billed separately and does not draw from the weekly Cursor-model pool. · Free tier: Yes: ongoing Free ($0) — limited Agent, Chat, and Tab with Cursor models (Grok and Composer); not a trial
In one agent turn with no follow-ups, Cursor built our T1 todo app as a single index.html in about 2.5 minutes and tested it in its built-in browser. We verified it ourselves in headless Chrome: add, complete, delete, filters, reload persistence and a 375px mobile layout all passed, and the only console error was a missing favicon. On Linux it needed --no-sandbox and showed a keyring dialog. Single run. desktop app, free plan
- Outcome
- 9
- Speed
- 8
- Cost
- 9
- DX
- 7
- Reliability
- 8
- Working todo app in one turn, about 2.5 minutes, with no follow-ups
- Every T1 check passed in our own headless browser test, including mobile
- Tested its own work in a built-in browser
- Free Hobby plan was enough for one T1; no visible quota limit hit (Agent Stats 0/394)
- On Linux it needed --no-sandbox and showed an OS keyring dialog on first launch
- The agent ran commands without asking for approval by default
- Free usage is published as Limited, not a hard public number; heavier Agent work needs a paid X Subscription plan
- Free and Plus do not include third-party models; Premium and above bill them at API list price from prepaid credit
Who it’s for: Developers who want an AI-first VS Code-style editor with a built-in agent on Cursor models. Use Free to evaluate Agent on those models. Plus adds a larger weekly Cursor-model pool and still has no third-party models and no own API keys. Premium and above is where the official docs put third-party models, Cloud Agents, Bugbot, and Origin.
7. Aider — 9.3/10
Open-source AI pair programming in the terminal. Tool details
Pricing: Free open source (Apache 2.0); pay your LLM provider · Free tier: Yes: free tool, BYO API key or local models
3/3 runs fixed our seeded cart bug (4/4 tests pass) in 9-13s with a free OpenRouter model, without touching the test file. Install needed uv because pip was blocked on the system Python. CLI, free model
- Outcome
- 9
- Speed
- 10
- Cost
- 10
- DX
- 8
- Reliability
- 10
- Fixed the seeded bug correctly in 3 of 3 runs, in 9-13 seconds
- Works with free OpenRouter models, so our test cost $0
- Built-in --auto-test loop runs your tests after each edit
- pip install was refused on Debian's system Python; we needed uv
- Unattended runs need several flags (--yes-always, --message, --test-cmd)
- Free model slugs change; one we tried returned 'unavailable for free'
Who it’s for: Terminal-first developers who want an open-source pair programmer that works with any model, including free ones.
8. Claude Code — 9.3/10
Anthropic's agentic coding tool for the terminal, IDE, desktop, and browser. Tool details
Pricing: Included in Claude Pro $20/mo ($17/mo annual, $200 upfront); Max from $100/mo (5x or 20x); or API pay-as-you-go · Free tier: No (not included in Claude Free)
Anthropic Sonnet 5.5 edition (headline): Claude Code CLI fixed our seeded cart bug in 3 of 3 runs (5.7–7.1s, 5 turns each). Test file untouched. Real API spend ~$0.035 for smoke + 3 runs at Sonnet 5.5 list prices ($2/$10 per MTok). Overall 9.3 (outcome 10, speed 10, cost 7, DX 8, reliability 10). Alongside: OpenRouter free-model edition the same day also scored 9.3 at $0 (cohere/north-mini-code:free; slower, free-tier cost 10) — docs/tests/claude-code/RESULTS.md. Opus not run. Anthropic Sonnet 5.5 API
- Outcome
- 10
- Speed
- 10
- Cost
- 7
- DX
- 8
- Reliability
- 10
- Fixed the seeded bug in 3/3 runs with Anthropic Sonnet 5.5 in about 6 seconds
- Real API cost about $0.01 per T3 run at Sonnet 5.5 list prices
- Did not modify the test file in any run
- OpenRouter free edition also available at $0 for CLI-loop testing (also 9.3)
- No Claude Free plan access — Anthropic models need Pro, Max, or API credits
- Cost score 7 reflects pay-as-you-go API, not a free tier
- Needs --dangerously-skip-permissions for unattended sandbox runs
- Opus edition not tested (Sonnet already maxed Outcome on this T3)
Who it’s for: Developers who want Anthropic's terminal agent on Sonnet/Opus (Pro, Max, or API). Headline score is Anthropic Sonnet 5.5 via API. A separate OpenRouter free edition (also 9.3) tests the CLI with a small free model only.
9. Cline — 9.2/10
Open-source autonomous coding agent for VS Code. Tool details
Pricing: Extension free; Teams $20/user/mo (first 10 seats free); Enterprise custom · Free tier: Yes: free for individuals, pay inference only
3/3 headless CLI runs fixed our seeded cart bug (4/4 tests pass) in 24-35s with a free OpenRouter model, without editing the test file. CLI, free model
- Outcome
- 9
- Speed
- 9
- Cost
- 10
- DX
- 8
- Reliability
- 10
- Headless CLI with a -y mode ran start to finish with no prompts
- Fixed the seeded bug in 3 of 3 runs and never touched the test file
- Works with free OpenRouter models, so our test cost $0
- Slower than Aider on the same model (24-35s vs 9-13s)
- Warns on Node versions below 22.15
- The -k flag saves your API key into the config directory
Who it’s for: Developers who like Cline in VS Code and want the same agent in a terminal or CI, with any model.
Related compares
- Cursor vs Claude Code
- Cursor vs Devin Desktop (formerly Windsurf)
- Cursor vs GitHub Copilot
- Lovable vs Bolt.new
- Lovable vs v0
- Claude Code vs OpenAI Codex
- Cursor vs Kiro
- Kiro vs GitHub Copilot
- Kiro vs Devin Desktop (formerly Windsurf)
- GitHub Copilot vs Devin Desktop (formerly Windsurf)
- v0 vs Bolt.new
- Bolt.new vs Base44
- Base44 vs Lovable
- Bolt.new vs Replit Agent
- v0 vs Replit Agent
- Aider vs Cline
- Cline vs OpenCode
- Aider vs OpenCode
- Cline vs OpenAI Codex
- Jules vs Gemini CLI
- Lovable vs Replit Agent
- Cursor vs Zed
- Kiro vs Zed
- GitHub Copilot vs Zed
- Aider vs OpenAI Codex
- OpenCode vs OpenAI Codex
- v0 vs Base44
- Lovable vs Google AI Studio
- Bolt.new vs Google AI Studio
- GitHub Copilot CLI vs Aider
- GitHub Copilot CLI vs OpenCode
- GitHub Copilot CLI vs OpenAI Codex
- GitHub Copilot CLI vs Cline
- v0 vs Google AI Studio
- Base44 vs Replit Agent
- Base44 vs Google AI Studio
- Replit Agent vs Google AI Studio
- OpenAI Codex vs Gemini CLI
- Jules vs OpenAI Codex
- Aider vs Gemini CLI
- Cline vs Gemini CLI
- Aider vs Reason
- Cline vs Reason
- OpenCode vs Reason
- OpenHands vs Claude Code
- Amp vs Claude Code
- TRAE vs OpenAI Codex
- Memdoor vs Aider
FAQ
What is the best Vibe Coding Tools in our tests?
Kiro has the highest recorded overall score among tools on this page, 9.7/10, but this vibe-coding guide lists app builders first rather than opening with that score. The score is for the task and edition recorded for that tool, not a purchase ranking.
Are rankings paid?
No. Scores come from our own tests and are never for sale. See our disclosure.
How fresh is this ranking?
The latest recorded check or test date on this page is 2026-11-12.
Can Kiro use Claude Opus and GPT-5.6 cheaply?
On paid plans, yes relative to buying each vendor's own top subscription: Pro is $20/mo for 1,000 credits and can select Claude Opus 5.5 (2.0× credits), Claude Sonnet 5.5 (1.3×), and GPT-5.6 Sol/Terra/Luna (4.4×/2.2×/0.6× vs Auto), per kiro.dev/docs/models checked 2026-11-02. Free cannot use those flagships — only Auto, Sonnet 4.5/4.0, and listed open-weight models.
What models does Kiro Free include?
Auto, Claude Sonnet 4.5, Claude Sonnet 4.0, DeepSeek 3.2, MiniMax M2.5, GLM-5, MiniMax M2.1, and Qwen3 Coder Next, with Free's 50 monthly credits and rate limits (models page, checked 2026-11-02).
Why aren’t the highest overall scores at the top?
This guide follows the prompt-to-app path, so app builders come first even when an editor or agent has a higher recorded overall score. Those scores are still shown inside each section. For a raw cross-category scoreboard, see Best AI Coding Tools.
Best free Vibe Coding Tools · All rankings · All tools · Best AI Coding Agents · Best AI Coding Tools
Scores come from our own tests and are never for sale; sponsored placements are always labelled. Some links are affiliate links (currently Railway). Disclosure