Ten tools that turn a description into software, scored the same way. App builders (Lovable, Bolt, v0, Replit Agent, Base44) give you a running, hosted app from a prompt. Coding agents (Cursor, Claude Code, OpenAI Codex, GitHub Copilot, Windsurf, now Devin Desktop) work inside a real code repository. They overlap, but they answer different questions: how fast can I get something in front of people, and how far can this go once it's real.
Every score uses the same six criteria, published on the methodology page, and comes from each vendor's own pricing pages and documentation, opened on November 1, 2026. None of these products is ours, and none of them pays us. Our own product, Skillspace, is a skills directory and isn't in this table.
| OpenAI CodexCoding agent | 3.8 | 3 | 5 | 2 | 5 | 4 | 4 | $20/mo (ChatGPT Plus) |
|---|---|---|---|---|---|---|---|---|
| Claude CodeCoding agent | 3.7 | 3 | 5 | 2 | 5 | 3 | 4 | $20/mo (Pro) |
| LovableApp builder | 3.7 | 5 | 3 | 4 | 3 | 3 | 4 | $25/mo (Pro, 100 credits) |
| Replit AgentApp builder | 3.7 | 4 | 4 | 5 | 3 | 2 | 4 | $20/mo (Core) |
| CursorCoding agent | 3.5 | 3 | 5 | 2 | 5 | 2 | 4 | $30/mo (Premium) |
| GitHub CopilotCoding agent | 3.5 | 2 | 3 | 2 | 5 | 4 | 5 | $10/mo (Pro) |
| v0App builder | 3.5 | 4 | 3 | 3 | 4 | 3 | 4 | $30/user/mo (Plus) |
| BoltApp builder | 3.3 | 5 | 3 | 3 | 3 | 3 | 3 | $25/mo (Pro, 10M tokens) |
| Windsurf (Devin Desktop)Coding agent | 3.3 | 3 | 4 | 2 | 5 | 3 | 3 | $20/mo (Pro) |
| Base44App builder | 3.0 | 5 | 2 | 4 | 2 | 3 | 2 | $16/mo billed annually (Starter) |
Scores out of 5 against our published rubric, from vendor docs and pricing pages. Hover or open a review for the reason behind each score. How we score
How to read the scores
A 5 in one column isn't better than a 5 in another; what matters is which columns you care about. A founder testing an idea should weight speed and backend; a team with a codebase should weight depth, exit and team. The overall number is a plain average, so it hides exactly the trade-offs that should decide your pick. That's why every review opens with "pick it if" and "skip it if" rather than the score.
Two patterns stand out in the table. App builders score high on speed and backend and lower on exit, because their convenience comes from hosting your data. Coding agents are the reverse: everything stays in your repository, and nothing is set up for you. If you're unsure which group you belong in, the decision tool asks eight questions and tells you whether it should be an app at all, and our app builders comparison walks through who should pick which. Two rankings re-weight the same scores for one job each: non-technical founders and internal tools. To price a month of building, use the build-cost calculator.
All reviews
Head-to-heads
The pairs people search for most, each with a side-by-side table built from the same data: Lovable vs Bolt, v0 vs Lovable, Replit vs Lovable, Base44 vs Lovable, Bolt vs v0, Cursor vs Claude Code, Claude Code vs Codex and Cursor vs Windsurf. All of them are on the compare page.
Questions people ask
Which AI app builder scores highest?
On our rubric, Lovable and Replit Agent tie among the app builders at 3.7 out of 5, from opposite strengths. Lovable is faster to a polished app with a more predictable bill; Replit gives more infrastructure and depth with effort-based pricing. Among the coding agents, OpenAI Codex scores 3.8, Claude Code 3.7, and Cursor and GitHub Copilot 3.5. Codex edges Claude Code only on pricing predictability, because OpenAI publishes per-model limits. Mahmoud uses Claude Code regularly, which the review discloses.
Did you test these tools?
Mostly we compared published pricing, plan pages and documentation, opened on November 1, 2026, and each review says so. Mahmoud has built with Lovable and uses Claude Code regularly, and those two reviews say that too. We don't run synthetic benchmarks, and we don't publish scores from other people's tests.
Do you get paid by any of these companies?
No. We take no affiliate commissions, sponsorships or payment for placement, and no link on these pages pays us. None of the ten products is ours.