---
title: "Vercel AI SDK Downloads vs. the Lab SDKs"
date: 2026-09-19
url: https://modevc.com/charts/vercel-ai-sdk-downloads-vs-lab-sdks/
keywords: ["vercel ai sdk downloads", "vercel ai sdk vs openai sdk", "ai sdk npm"]
publisher: Mode VC
---

# Vercel AI SDK Downloads vs. the Lab SDKs

Vercel's AI SDK grew 6.5× in under a year, but the lab SDKs grew 8.7×. The layer that makes models swappable did not capture the growth.

## Chart: The Vercel AI SDK fell from 31 to 23 weekly downloads per 100 downloads of the OpenAI, Anthropic and Google SDKs, after a peak of 40.

Weekly npm downloads of the ai package per 100 combined downloads of openai, @anthropic-ai/sdk and @google/genai. Complete Monday–Sunday weeks, 6 Oct 2025 – 30 Aug 2026.

| Week of | ai | Lab SDKs combined | ai per 100 lab SDK |
|---|---|---|---|
| 6 Oct 2025 | 3.6M | 11.7M | 31.1 |
| 13 Oct 2025 | 3.7M | 12.2M | 30.8 |
| 20 Oct 2025 | 4.0M | 11.7M | 34.1 |
| 27 Oct 2025 | 4.2M | 11.6M | 36.4 |
| 3 Nov 2025 | 4.6M | 12.2M | 37.5 |
| 10 Nov 2025 | 4.8M | 12.3M | 39.2 |
| 17 Nov 2025 | 5.1M | 13.3M | 38.5 |
| 24 Nov 2025 | 4.9M | 12.1M | 40.4 |
| 1 Dec 2025 | 5.5M | 14.4M | 38.0 |
| 8 Dec 2025 | 5.2M | 14.5M | 35.9 |
| 15 Dec 2025 | 5.3M | 14.4M | 36.5 |
| 22 Dec 2025 | 3.1M | 9.4M | 32.4 |
| 29 Dec 2025 | 2.8M | 9.6M | 29.6 |
| 5 Jan 2026 | 5.6M | 15.7M | 35.5 |
| 12 Jan 2026 | 6.2M | 16.7M | 37.0 |
| 19 Jan 2026 | 6.4M | 18.2M | 35.1 |
| 26 Jan 2026 | 7.0M | 21.8M | 32.3 |
| 2 Feb 2026 | 7.6M | 22.8M | 33.2 |
| 9 Feb 2026 | 7.8M | 23.9M | 32.6 |
| 16 Feb 2026 | 8.0M | 23.4M | 34.0 |
| 23 Feb 2026 | 8.7M | 28.8M | 30.2 |
| 2 Mar 2026 | 9.9M | 31.4M | 31.5 |
| 9 Mar 2026 | 10.1M | 35.4M | 28.5 |
| 16 Mar 2026 | 9.8M | 34.9M | 28.1 |
| 23 Mar 2026 | 10.4M | 35.9M | 29.1 |
| 30 Mar 2026 | 10.1M | 37.1M | 27.2 |
| 6 Apr 2026 | 10.6M | 40.1M | 26.5 |
| 13 Apr 2026 | 11.5M | 42.9M | 26.9 |
| 20 Apr 2026 | 11.9M | 45.3M | 26.2 |
| 27 Apr 2026 | 12.1M | 50.7M | 23.9 |
| 4 May 2026 | 12.6M | 48.0M | 26.2 |
| 11 May 2026 | 13.2M | 50.4M | 26.3 |
| 18 May 2026 | 13.7M | 54.0M | 25.5 |
| 25 May 2026 | 13.0M | 61.9M | 21.0 |
| 8 Jun 2026 | 15.1M | 65.2M | 23.2 |
| 15 Jun 2026 | 15.7M | 68.8M | 22.9 |
| 29 Jun 2026 | 15.1M | 63.7M | 23.8 |
| 13 Jul 2026 | 17.7M | 72.7M | 24.4 |
| 20 Jul 2026 | 18.8M | 74.7M | 25.1 |
| 27 Jul 2026 | 19.8M | 80.2M | 24.7 |
| 3 Aug 2026 | 20.6M | 79.3M | 25.9 |
| 17 Aug 2026 | 22.4M | 89.6M | 25.0 |
| 24 Aug 2026 | 23.6M | 101.8M | 23.2 |

Note: ai is the core package of Vercel's AI SDK; model providers plug in through adapter packages such as @ai-sdk/openai. Downloads count installs, including CI and packages that depend on these SDKs, so lab SDK counts include other tools built on top of them. Weeks with an npm gap day (the API returns 0) are dropped for all packages; the latest complete week is 24 Aug.

## The signal

- **Launches:** 10 model launches from 8 labs in our launch feed between 1 and 19 September; every one is a reason to swap models.
- **npm:** Inside the AI SDK, the Anthropic adapter went from 61% to 100% of the OpenAI adapter's weekly downloads (11.9M vs 11.9M, week of 24 Aug).
- **npm:** ai had 23.6M downloads in the week of 24 Aug, up from 3.6M in the week of 6 Oct 2025.

## The Mode read

Our launch feed lists 10 new models from 8 labs in the first 19 days of September. A layer above the models, such as Vercel's AI SDK, promises that a team can try each one by changing a line of config. If models really are swappable, that layer should be taking a growing share of developer installs. Is it?

Not on npm. The ai package grew **6.5×**, from 3.6M to 23.6M weekly downloads. The three lab SDKs together grew **8.7×**. Per 100 lab SDK downloads, ai went from 31 to a peak of 40 in the week of 24 Nov 2025, then down to 23. Swapping does happen inside the layer: the Anthropic adapter went from 61% to 100% of the OpenAI adapter.

So the abstraction works, and it still did not take the market. Most new installs went straight to a lab. A neutral layer makes the model a commodity, which is our thesis, but making something a commodity does not mean you own the customer. The switch is useful. The moat is whatever the developer cannot move as easily: their deployment, their data, their evals, their bill.

> Making models swappable is a feature. Owning what cannot be swapped is the business.

**What we would back:** Developer platforms that pair a model-neutral layer with something sticky: hosting, data, evals or billing.

## Sources

- [npm registry downloads API](https://api.npmjs.org/downloads/range/2025-10-01:2026-10-06/ai)
- [Official launch posts (launch feed, captured 7 Oct 2026)](https://openai.com/index/gpt-6-astra/)