Quick answer: what are the fastest-growing npm packages in 2026?
The fastest-growing npm packages in this PkgPulse refresh are rolldown, hono, @tanstack/react-router, oxc-transform, @anthropic-ai/sdk, effect, ai, valibot, drizzle-orm, and elysia. They are growing because three adoption waves are overlapping: Rust-based JavaScript tooling, edge/server-first TypeScript frameworks, and AI SDKs moving from experiments into production apps.
If you searched for the most popular or most downloaded npm packages in 2026, the answer is different. Mature baselines such as typescript, zod, eslint, react, vite, prettier, and express still generate much larger weekly download totals than most emerging packages. Growth rate tells you which packages are gaining share fastest; weekly downloads tell you which packages are already entrenched.
This guide preserves that distinction. It ranks growth signals first, then shows the most-downloaded context so you can tell whether a package is a breakout candidate, a mainstream default, or both.
2026 npm package count and download context
As of May 15, 2026, the public npm registry replication endpoint reported about 4.04 million registry documents (doc_count: 4,038,347). Treat that as a registry-scale indicator, not a perfect count of actively maintained production packages: package docs can include abandoned, deprecated, scoped, and very low-usage packages.
For downloads, this refresh uses npm's public download API for the week 2026-05-08 through 2026-05-14, compared with the same calendar week in 2025. The most-downloaded packages in the measurement set were:
| Package | Weekly downloads, 2026-05-08..14 | Same week 2025 | Same-week growth |
|---|---|---|---|
typescript | 201.5M | 83.2M | 142% |
zod | 142.2M | 30.9M | 360% |
eslint | 133.6M | 55.6M | 140% |
react | 132.5M | 37.9M | 250% |
vite | 125.4M | 27.7M | 353% |
prettier | 106.3M | 50.9M | 109% |
express | 103.5M | 43.8M | 136% |
webpack | 48.2M | 31.0M | 56% |
vitest | 45.1M | 11.0M | 311% |
next | 35.8M | 10.2M | 252% |
That table is why a pure "top npm packages by downloads" list would look conservative: the highest-volume packages are mostly ecosystem foundations. The ranking below is more useful when you want to spot fast adoption changes before they become obvious from raw popularity alone.
Methodology and caveats
- Primary source: npm public downloads API, using
https://api.npmjs.org/downloads/point/2026-05-08:2026-05-14/<package>and the same endpoint for2025-05-08:2025-05-14. - Registry scale source: npm public replication endpoint at
https://replicate.npmjs.com/, accessed May 15, 2026. - Ranking rule: same-calendar-week year-over-year download growth, rounded for readability.
- Scope: package-health candidates that are relevant to JavaScript application teams, PkgPulse comparison pages, and 2026 stack decisions. This is a reproducible editorial ranking, not an official npm global leaderboard.
- Noise warning: npm downloads are not installs or active users. CI systems, lockfile churn, monorepo package managers, registry caches, and release trains can inflate or shift weekly counts. Use the numbers as directional adoption signals, not precise usage telemetry.
- Threshold note: most packages cleared a meaningful 2025 base before inclusion.
rolldownsits right around the threshold and is called out because the Vite/Rolldown/Oxc transition is strategically important for the JavaScript build-tool ecosystem.
Top 20 fastest-growing npm packages in 2026
| Rank | Package | Weekly downloads | Same week 2025 | Growth | What the signal means |
|---|---|---|---|---|---|
| 1 | rolldown | 28.8M | 50K | 57,956% | Rust-based Rollup-compatible bundling moved from future roadmap to practical ecosystem dependency. |
| 2 | hono | 33.6M | 958K | 3,409% | Edge and multi-runtime server apps are choosing Hono as the lightweight default. |
| 3 | @tanstack/react-router | 16.0M | 483K | 3,201% | Type-safe routing demand is moving beyond niche TanStack users. |
| 4 | oxc-transform | 3.4M | 106K | 3,144% | Oxc is becoming real infrastructure behind faster JavaScript toolchains. |
| 5 | @anthropic-ai/sdk | 17.9M | 1.3M | 1,275% | Claude API usage has crossed from experiments into production developer workflows. |
| 6 | effect | 14.5M | 1.3M | 1,014% | TypeScript teams are adopting typed error, concurrency, and dependency patterns at larger scale. |
| 7 | ai | 13.1M | 1.2M | 981% | The Vercel AI SDK became a default abstraction for streaming and multi-provider AI features. |
| 8 | valibot | 10.8M | 1.3M | 713% | Bundle-size pressure is creating room for Zod alternatives. |
| 9 | drizzle-orm | 9.3M | 1.2M | 648% | SQL-first TypeScript ORM adoption keeps compounding in serverless and edge stacks. |
| 10 | elysia | 480K | 71K | 574% | Bun-native frameworks remain smaller than Hono, but their adoption curve is steep. |
| 11 | bun | 1.6M | 246K | 555% | The npm package is only a partial proxy for Bun runtime use, but it shows sustained ecosystem pull. |
| 12 | openai | 19.7M | 3.9M | 406% | AI provider SDKs are now mainstream application dependencies. |
| 13 | @biomejs/biome | 8.8M | 1.9M | 376% | Fast linting/formatting is no longer just an early-adopter concern. |
| 14 | zod | 142.2M | 30.9M | 360% | Zod remains both high-growth and high-volume despite smaller validators gaining ground. |
| 15 | vite | 125.4M | 27.7M | 353% | Vite is still the build-tool gravity well, not merely a mature baseline. |
| 16 | @rspack/core | 6.1M | 1.4M | 343% | Webpack-compatible speedups are attracting teams that cannot rewrite their build configs. |
| 17 | arktype | 947K | 229K | 315% | TypeScript-native validation is getting enough usage to matter in stack choices. |
| 18 | vitest | 45.1M | 11.0M | 311% | Vite-native testing continues to pull demand away from older Jest-first defaults. |
| 19 | @trpc/server | 3.4M | 928K | 270% | End-to-end type-safe APIs remain a durable pattern, especially in TypeScript-first apps. |
| 20 | next | 35.8M | 10.2M | 252% | Next.js is already mainstream, but the package still shows strong same-week growth. |
What is driving the growth?
1. Rust-based JavaScript tooling is now mainstream infrastructure
rolldown, oxc-transform, @biomejs/biome, and @rspack/core all point in the same direction: JavaScript teams want faster feedback loops without abandoning familiar configuration models. Rspack is compelling because it can meet webpack users where they are. Biome is compelling because it replaces multiple lint/format steps with one fast tool. Oxc and Rolldown matter because they sit beneath the next generation of Vite-compatible bundling.
The practical implication: if your team still treats build speed as a fixed cost of using JavaScript, 2026 is the year to re-evaluate. Start with the migration surfaces that minimize rewrite risk: Vite vs Webpack, Biome vs Oxc, Oxc vs SWC, and Rolldown vs Rspack vs Turbopack.
2. Edge and server-first packages are becoming defaults
hono, drizzle-orm, elysia, @trpc/server, and next are not growing because every team suddenly wants more framework surface area. They are growing because app architecture shifted toward server actions, server components, edge functions, and serverless databases. Packages that make that world easier are benefiting.
Hono is the clearest breakout: it runs across Cloudflare Workers, Bun, Deno, and Node.js, and its download growth now dwarfs what a niche edge framework would look like. Drizzle is the database-side counterpart: it gives TypeScript teams SQL-first schema control while still fitting serverless deployment patterns. For adjacent decisions, see Hono vs Express vs Fastify vs Elysia, Kysely vs Drizzle vs Prisma, and Bun vs Deno 2 vs Node.
3. AI SDKs are now ordinary app dependencies
@anthropic-ai/sdk, ai, and openai are high on the growth table because AI features moved from prototypes into product surfaces. The important nuance is that these packages are not interchangeable signals. The provider SDKs (openai, @anthropic-ai/sdk) show direct model-provider adoption. The ai package shows demand for a higher-level streaming and provider-abstraction layer.
That matters for architecture. If you expect provider churn, the Vercel AI SDK style of abstraction can reduce switching cost. If your application depends on provider-specific capabilities, direct SDK usage may still be the right interface. For deeper stack choices, see AI SDK vs LangChain, AI SDK vs OpenAI Agents SDK vs LangChain, and how AI is changing npm package choice.
4. Validation is splitting into high-volume defaults and lean challengers
zod is the surprise here if you only follow social chatter. It is not being replaced wholesale; it is still one of the highest-volume packages in the set and grew about 360% in the same-week comparison. At the same time, valibot and arktype show that developers are seriously evaluating smaller, faster, or more TypeScript-native validators.
The decision is not simply "Zod versus everything else." For server-only validation, Zod's ecosystem and familiarity can matter more than bundle size. For client bundles, edge runtimes, or high-frequency parsing, Valibot and ArkType deserve a fresh look. See Zod vs Yup vs Joi, Zod vs Yup TypeScript validation, and smallest bundle npm packages under 5KB.
5. Type-safe app structure is still compounding
@tanstack/react-router, @trpc/server, @t3-oss/env-nextjs just outside the top 20, and effect all point toward the same buyer intent: TypeScript teams want mistakes caught at build time. The growth is not limited to one framework or one architectural school. It shows up in routing, APIs, environment variables, effects, and validation.
This is also where download counts can overstate certainty. A package can spike because a popular starter template or monorepo scaffold pulls it in. The stronger signal is when the package also has visible documentation maturity, active releases, and multiple independent frameworks adopting the pattern. That is why PkgPulse treats downloads as one input alongside maintenance and ecosystem fit.
Which npm packages are most popular in 2026?
If your question is popularity rather than growth, the answer is still dominated by foundations:
typescript,react,eslint,prettier,vite,express, andwebpackremain enormous by weekly downloads.zodis both popular and still fast-growing, which makes it an unusually strong adoption signal.vitestandnextnow sit in the category where they are too large to call niche but still growing quickly enough to watch.hono,ai,@anthropic-ai/sdk,openai, and@tanstack/react-routerare the clearest examples of packages moving from specialist awareness to mainstream dependency graphs.
For stack decisions, do not choose solely from the popularity list. A high-volume package can be the safest default, but a high-growth package can be the better bet when it solves a specific pain point in your architecture.
What to watch through the rest of 2026
First, watch the Rolldown/Oxc/Vite path. If Vite's Rust-backed internals keep moving forward, Oxc-related packages can grow without every developer explicitly choosing Oxc. That kind of transitive adoption can produce very large download jumps.
Second, watch edge-first server libraries. Hono's growth suggests developers are treating multi-runtime compatibility as a core requirement, not an optimization. Elysia's smaller base makes it more volatile, but its Bun-native pitch remains aligned with the same trend.
Third, watch AI SDK layering. If product teams standardize on provider abstraction, ai keeps benefiting. If teams need provider-specific features, openai and @anthropic-ai/sdk keep growing directly. Many production apps will use both layers.
Fourth, watch validation and type-safe structure. The ecosystem has not settled on a single winner. Zod's scale, Valibot's bundle story, ArkType's TypeScript-native design, Effect's system-level approach, TanStack Router's route typing, and tRPC's API typing are all different answers to the same pressure: fewer runtime surprises.
Source notes
Accessed May 15, 2026:
- npm downloads API:
https://api.npmjs.org/downloads/point/2026-05-08:2026-05-14/<package>andhttps://api.npmjs.org/downloads/point/2025-05-08:2025-05-14/<package>. - npm public registry replication metadata:
https://replicate.npmjs.com/. - Package metadata and current weekly download cross-checks from npm registry search/package metadata.
- Project documentation consulted for positioning and ecosystem context: Biome, Oxc, Rolldown, Rspack, Hono, TanStack Router, Drizzle, Effect, and AI SDK.
The growth percentages in this article are rounded and should be rechecked before citing exact numbers in investor decks, procurement memos, or benchmark comparisons. They are strong enough for trend analysis, not a replacement for your own dependency-risk review.
How to use this ranking
Use the growth table as a shortlist, not as a mandate. A package that grows 1,000% from a smaller base can still be riskier than a slower-growing default with deeper maintenance history. A package with 100M weekly downloads can still be the wrong choice if it does not fit your runtime or bundle constraints.
The best workflow is:
- Start with the growth signal in this guide.
- Compare the package against your actual alternatives.
- Check maintenance, release cadence, type support, runtime support, and migration cost.
- Pilot the package in a bounded slice before putting it on a critical path.
For download trends, version history, and package-health signals, start with PkgPulse comparison pages such as Vite vs Webpack, then move to the package-specific guides linked above.
See also: The 50 Most Underrated npm Packages in 2026, The 20 npm Packages Losing Downloads the Fastest 2026, and npm download trends: which frameworks are actually growing.