Which library should I use: Zod or Valibot?

Use Zod if ecosystem maturity, community support, and first-class integrations are your priority. Use Valibot if you are building for edge runtimes, browser bundles, or performance-critical client-side code and bundle size is a top concern. Both are TypeScript-first with automatic type inference.

Zod vs Valibot

Q: What is the main difference between Zod and Valibot?

The key architectural difference is tree-shakeability. Zod ships a monolithic bundle (~13 KB min) where importing any validator loads the whole library. Valibot is modular — each validator is a separate export, so bundlers only include what you actually import. A minimal Valibot schema can be under 1 KB after tree-shaking. Both are TypeScript-first with automatic type inference, and their APIs are deliberately similar.

Q: Is Valibot faster than Zod?

Benchmarks show Valibot is generally faster than Zod for validation throughput, often by 2-5x on simple schemas. This is partly because Valibot's modular design avoids some overhead present in Zod's monolithic parser. However, for most applications the difference is imperceptible in practice — both are well under 1ms per validation on typical schemas.

Q: Does Valibot have the same API as Zod?

The APIs are similar but not identical. Valibot uses pipe() to chain validators instead of method chaining (.min(), .max() on the schema itself). For example, Zod's z.string().min(2).email() becomes v.pipe(v.string(), v.minLength(2), v.email()) in Valibot. Both infer TypeScript types the same way (v.InferOutput vs z.infer ).

Q: Is Valibot production-ready?

Valibot is production-ready and actively maintained, but its ecosystem is younger than Zod's. Zod has been the de facto standard since 2020 with first-class integrations in tRPC, React Hook Form, Astro, and hundreds of other libraries. Valibot has adapters for most major integrations but you may encounter edge cases not yet covered by the community.

Q: Should I migrate from Zod to Valibot?

Only migrate if bundle size is a critical constraint — for example, edge functions or browser code where every KB matters. If your app is server-rendered or bundle size is not a bottleneck, the migration cost (rewriting schemas, updating integrations) is unlikely to be worth the savings. New projects where bundle size is a concern should evaluate Valibot from the start.

Both are TypeScript-first schema validators. The difference is architecture: Zod is battle-tested and monolithic; Valibot is tree-shakeable and ultra-lightweight.

At a Glance

Feature	Zod	Valibot
TypeScript-first	Yes	Yes
Type inference	z.infer<typeof schema>	v.InferOutput<typeof schema>
Bundle size (min)	~13 KB (monolithic)	< 1 KB per validator (tree-shakeable)
Architecture	Monolithic — import loads all	Modular — import only what you use
API style	Method chaining (.string().min(2))	pipe() composition (pipe(string(), minLength(2)))
Validation speed	Fast	Generally faster (2-5x on benchmarks)
Ecosystem maturity	Very high — de facto standard	Growing — younger community
tRPC integration	First-class	Adapter available
React Hook Form	Official resolver	Valibot resolver available
Async validation	Yes (parseAsync)	Yes (parseAsync)
Error formatting	ZodError with flatten()	ValiError with structured issues
Best for	Most TypeScript projects	Edge/browser — bundle size critical

The Core Difference: Bundle Size & Architecture

Zod bundles everything together. When you write import { z } from "zod", the whole ~13 KB library loads — even validators you never use. For server-side Node.js, this is irrelevant. For edge functions (Cloudflare Workers, Vercel Edge) or browser bundles, every kilobyte counts.

Valibot is built from the ground up as individual functions that bundlers can tree-shake. If your schema only uses string, email, and object, only those few hundred bytes land in your bundle.

Example: A single email validation schema in Zod costs ~13 KB. The same schema in Valibot costs roughly 600 bytes — about 20x smaller. For a Cloudflare Worker with a 1 MB size limit, this matters.

Same Schema, Two Libraries

User registration schema

Zod

import { z } from "zod";

const RegisterSchema = z.object({
  username: z.string()
    .min(3)
    .max(20)
    .regex(/^[a-z0-9_]+$/),
  email: z.string().email(),
  password: z.string().min(8),
  age: z.number().int().min(13).optional(),
});

type RegisterInput = z.infer<typeof RegisterSchema>;

// Parse (throws on error):
const data = RegisterSchema.parse(input);

// Safe parse (no throw):
const result = RegisterSchema.safeParse(input);
if (result.success) {
  console.log(result.data.username);
}

Valibot

import * as v from "valibot";

const RegisterSchema = v.object({
  username: v.pipe(
    v.string(),
    v.minLength(3),
    v.maxLength(20),
    v.regex(/^[a-z0-9_]+$/)
  ),
  email: v.pipe(v.string(), v.email()),
  password: v.pipe(v.string(), v.minLength(8)),
  age: v.optional(v.pipe(v.number(), v.integer(), v.minValue(13))),
});

type RegisterInput = v.InferOutput<typeof RegisterSchema>;

// Parse (throws on error):
const data = v.parse(RegisterSchema, input);

// Safe parse (no throw):
const result = v.safeParse(RegisterSchema, input);
if (result.success) {
  console.log(result.output.username);
}

Discriminated unions

Zod

const EventSchema = z.discriminatedUnion("type", [
  z.object({
    type: z.literal("click"),
    x: z.number(),
    y: z.number(),
  }),
  z.object({
    type: z.literal("keydown"),
    key: z.string(),
  }),
]);

type Event = z.infer<typeof EventSchema>;

Valibot

const EventSchema = v.variant("type", [
  v.object({
    type: v.literal("click"),
    x: v.number(),
    y: v.number(),
  }),
  v.object({
    type: v.literal("keydown"),
    key: v.string(),
  }),
]);

type Event = v.InferOutput<typeof EventSchema>;

Choose Zod when:

✓ You want the largest ecosystem and most community resources
✓ Using tRPC (Zod is the native validator)
✓ Bundle size is not a primary constraint
✓ Your team already knows Zod
✓ You rely on third-party integrations (most target Zod first)
✓ Building a server-side Node.js or Next.js app

Choose Valibot when:

✓ Targeting edge runtimes (Cloudflare Workers, Vercel Edge)
✓ Building browser-side validation with a tight bundle budget
✓ Maximum validation performance is critical
✓ Starting a new project and want the modern modular approach
✓ Your framework/library already has Valibot adapters

Frequently Asked Questions

What is the main difference between Zod and Valibot?

The key difference is architecture. Zod is monolithic (~13 KB min) — any import loads the full library. Valibot is modular and tree-shakeable, so bundlers only include the validators you actually use. A minimal Valibot schema can be under 1 KB. Both are TypeScript-first with automatic type inference.

Is Valibot faster than Zod?

Benchmarks show Valibot is generally faster (2-5x on simple schemas). In practice, both are well under 1ms per validation for typical schemas, so the difference is usually imperceptible in application code.

Does Valibot have the same API as Zod?

The APIs are similar but not identical. Valibot uses pipe() to compose validators instead of method chaining. Zod's z.string().min(2).email() becomes v.pipe(v.string(), v.minLength(2), v.email()) in Valibot. Type inference uses v.InferOutput<typeof schema> instead of z.infer<typeof schema>.

Is Valibot production-ready?

Valibot is production-ready and actively maintained, but younger than Zod. Zod has first-class integrations in tRPC, React Hook Form, Astro, and many other libraries. Valibot has adapters for most but you may hit edges not yet covered by the community.

Should I migrate from Zod to Valibot?

Only migrate if bundle size is a critical constraint — edge functions or browser code where every KB matters. If your app is server-rendered or bundle size is not a bottleneck, the migration cost is unlikely to be worth it.

Which should I use for a new TypeScript project?

Use Zod for most projects — the ecosystem, community, and integrations are unmatched. Use Valibot if you are building for edge or browser with strict bundle size constraints from the start.

Working with Zod? The cheatsheet has everything.

All schema types, refinements, transforms, and error handling patterns in one page.

Zod Cheatsheet →

From the blog

View all →