// side by side · same rules, same day

v0 vs Bolt.new

v0 5/5 · Bolt.new 2/5

Download the card v0 dossier →Bolt.new dossier →

The three axes

	v0Vercel AGENT	Bolt.newStackBlitz WRAPPER WITH AMBITION
CAPABILITYyes / no / uncertain	5/5 AGENT	2/5 WRAPPER WITH AMBITION
PROOFhow well you can check us	DOCS CLAIMED	OPERATOR INSPECTABLE
TRACE ACCESSwho can read the record	VENDOR-ONLYSteps appear in a demo or worked example you cannot run	OPERATOR-OWNEDWhoever runs it holds the record; nothing is published
LEASHtool calls between human turns	not measured — measure it yourself	not measured — measure it yourself
CHECKEDversion and date	v0 (v0.app docs: agentic features, terminal commands)2026-08-11	Bolt (support.bolt.new docs, 2026)2026-08-11
CONFIDENCEours, in this dossier	7/10	6/10
COMMUNITYwhat visitors ruled	A0 · W0 · C0n=0 raw votes · signal at 25	A0 · W0 · C0n=0 raw votes · signal at 25
IN COURTopen challenges	no cases filed	no cases filed
DISPUTEDwhere the axes disagree	LOOPS — yes on the vendor's word alone CHOOSES — yes on the vendor's word alone ACTS — yes on the vendor's word alone RECOVERS — yes on the vendor's word alone UNSUPERVISED — yes on the vendor's word alone	nothing disputed — every answer is inspectable

Question by question

rows marked "they differ" are where the comparison earns its keep

	v0Vercel AGENT	Bolt.newStackBlitz WRAPPER WITH AMBITION
LOOPSruns until done	✓ yes DOCS CLAIMED confidence 8.5/10 · checked 2026-08-11 v0's live docs show a real agent loop: it coordinates multi-step workflows, has Auto-continue mode, opens/tests/critiques apps and fixes proactively, recovering and trying alternatives until done. Step count varies by task.	✓ yes OPERATOR INSPECTABLE confidence 7/10 · checked 2026-08-11 One prompt drives a variable-length run: the agent decides file edits, whether the app needs a database, which skills to load; the security audit fixes what it finds and flags the rest.
CHOOSESpicks its own toolsthey differ	✓ yes DOCS CLAIMED confidence 9/10 · checked 2026-08-11 Docs: v0 "will automatically use the appropriate capabilities based on what you're asking for" — web search, browser, terminal, MCP picked at runtime per request, not pre-wired. Published NO contradicts its own rationale.	✗ no OPERATOR INSPECTABLE confidence 6/10 · checked 2026-08-11 Only a Standard-vs-Max model choice is documented; runtime tool or MCP-connector selection is not.
ACTSchanges the world	✓ yes DOCS CLAIMED confidence 7/10 · checked 2026-08-11 Writes files and runs bash in its sandbox — unit tests, Vercel and GitHub CLI calls.	✓ yes OPERATOR INSPECTABLE confidence 9/10 · checked 2026-08-11 Writes and runs project files in a WebContainer, auto-commits to the connected GitHub repo, provisions a Bolt Cloud database, and via MCP connectors edits Notion pages or opens GitHub PRs.
RECOVERSfixes its own errorsthey differ	✓ yes DOCS CLAIMED confidence 9/10 · checked 2026-08-11 Docs confirm rationale: "Recovers from issues and tries alternative approaches" and it diagnoses/fixes errors "automatically as part of the generation loop" while progressing multi-step tasks without approval. Verdict was wrong.	✗ no OPERATOR INSPECTABLE confidence 6/10 · checked 2026-08-11 Troubleshooting doc tells the user to prompt 'The preview is not showing, fix this.' themselves.
UNSUPERVISEDnobody watchingthey differ	✓ yes DOCS CLAIMED confidence 9.5/10 · checked 2026-08-11 v0 docs name "Full" mode, which skips all terminal permission checks so a full task runs with no per-command approval; "Auto" default also runs allowlisted commands unprompted. Rule 05 satisfied.	✗ no OPERATOR INSPECTABLE confidence 6/10 · checked 2026-08-11 Lifecycle doc is prompt-per-change; Plan mode waits for the user's 'Implement plan' click.

Every answer here is the same one on the dossier and in the dataset — this page holds no numbers of its own. Think one is wrong? Bring a trace.