v0 (Vercel)
v0's own benchmark: 93.9% error-free code generation vs 78.4% for raw Claude Opus on the same task. That's a real edge for Next.js scaffolding, not a vibe-coding gimmick. But it's still credit-metered pricing with no hard spending cap, and chat history counts as input tokens on every turn, so cost per feature climbs as a session grows.
PRICING
| Free | $0/mo | $5 included credit, 7 messages/day |
| Plus | $30/user/mo | $30 credit/user + $2 daily bonus, team billing |
| Business | $100/user/mo | same included credits as Plus; training opt-out on by default |
| Enterprise | Custom | SAML SSO, RBAC, priority access, SLA support |
Model usage is metered per token on top of included credits (from $1/1M input & $5/1M output on the fastest tier up to $10/1M input & $50/1M output on v0 Max Fast). Chat history, source files and Vercel-specific context all count as input tokens on every message, so long sessions get progressively pricier per generation.
checked 2026-07-27 · vendor pricing page
Element scores
Strengths
512K-token context on v0-1.5-lg, a two-way MCP implementation (client and server), one-click Marketplace integrations for Neon/Supabase/Stripe with scoped permission modes, automated secret-scanning that blocked over 17,000 insecure deployments in one 30-day window, and platform-level SOC 2 Type 2 plus ISO 27001 compliance.
Honest dings
Chat history counts as input tokens on every message, so credit burn compounds as a session grows rather than staying flat per feature. No customer-facing eval or trace UI inside v0 itself, despite Vercel's own detailed internal eval practice. Fine-tuning, synthetic data generation, multi-agent orchestration and a first-party vector store are simply out of scope.
Sources (11) — every claim traceable
Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).
- v0.app/pricing — Official pricing tiers (Free/Plus/Business/Enterprise) and per-token model pricing (accessed 2026-07-27)
- v0.app/docs — Product overview, self-description, core capabilities (accessed 2026-07-27)
- v0.app/docs/MCP — Bring-your-own MCP servers, Marketplace integration tool calls, permission modes (accessed 2026-07-27)
- v0.app/docs/api/platform/adapters/mcp-server — v0 exposed as an MCP server for Cursor/Claude Desktop/VS Code (accessed 2026-07-27)
- vercel.com/blog/v0-composite-model-family — Composite model architecture, base models (Sonnet 4), error-free generation benchmark vs Claude Opus/Gemini/o3/GPT-4.1 (accessed 2026-07-27)
- vercel.com/blog/introducing-the-new-v0 — Official user count (4M+) and 2026 agentic-features relaunch (accessed 2026-07-27)
- v0.app/docs/security — Code analysis, sandboxed execution, SOC 2 Type 2 and GDPR compliance for v0 specifically (accessed 2026-07-27)
- vercel.com/blog/v0-vibe-coding-securely — 17k insecure deployments blocked in 30 days; deploy-time guardrails detail (accessed 2026-07-27)
- vercel.com/i/what-are-llm-evals-developers-primer — Vercel's own account of how they eval v0 internally (not a v0 product feature) (accessed 2026-07-27)
- vercel.com/docs/security/compliance — SOC 2 Type 2, ISO 27001, PCI DSS, HIPAA platform-level compliance detail (accessed 2026-07-27)
- chat.v0.dev/docs/api/model — v0-1.0-md/v0-1.5-md/v0-1.5-lg context window and output size specs (accessed 2026-07-27)