Sign in
← AI PulseAug 4, 2026

Wire · news · Multi-source brief

Vercel AI SDK Updates Baseten Integration for Embeddings

The Vercel AI SDK has updated its integration with Baseten, making the native performance client for embeddings an opt-in feature and removing it as a default dependency.

By Illumora Editorial

Source · Aug 4, 2026, 5:16 AM · On Illumora · Aug 4, 2026, 5:28 AM

Media from the primary source — shown here so you can stay on Illumora.

Synthesized from multiple allowlisted primaries on the same event. Lanes →

Brief drafted by Illumora’s editorial model from the linked primary source. Ops desk reviews flagged pieces. How we write →

Read the source →Vercel AI SDK Changelog — Release @ai-sdk/vue@4.0.51 · vercel/ai
Save

The Vercel AI SDK has released an update to its @ai-sdk/baseten package, version 2.1.0, which modifies how the native performance client for embeddings is handled. This change makes the client opt-in, removing it as a default dependency.

Key Points

  • The @ai-sdk/baseten@2.1.0 release makes the native performance client for embeddings opt-in.
  • The @basetenlabs/performance-client is no longer a default dependency of the SDK.
  • The removed dependency was a NAPI addon comprising 16 platform binary packages.
  • This addon typically required 5-16 MB of installed space.
  • The previous client could not load in edge runtimes and its platform binaries were not resolvable by bundlers.
  • Embeddings now use plain HTTP to communicate with Baseten's OpenAI-compatible endpoint.
  • Users can still install the native client and pass its constructor to createBaseten to retain client-side batching and request hedging.

Context

According to the Vercel AI SDK Changelog, the @basetenlabs/performance-client was previously imported at the module top level, meaning every consumer incurred its overhead, even if only embeddings utilized it. This client, being a NAPI addon, presented challenges with edge runtimes and bundler resolution due to its 16 platform binary packages and 5-16 MB installation size. The change aims to address these issues by allowing embeddings to go over plain HTTP to the deployment's OpenAI-compatible endpoint, which Baseten Embeddings Inference serves without additional settings.

Why It Matters

This update impacts developers using the Vercel AI SDK with Baseten, particularly those deploying to edge runtimes or seeking to reduce bundle sizes. By making the performance client opt-in, the SDK reduces default overhead and improves compatibility for certain deployment environments, while still offering the option for advanced features like client-side batching for those who require it.

What To Do

  • Note that @basetenlabs/performance-client is no longer automatically included with @ai-sdk/baseten.
  • If deploying to edge runtimes, verify that embeddings now function correctly via plain HTTP.
  • If client-side batching or request hedging are critical, install @basetenlabs/performance-client manually.
  • Review the createBaseten function documentation for how to pass the PerformanceClient constructor if needed.