Sign in
← AI PulseSep 15, 2026

Wire · news · Multi-source brief

Vercel AI SDK Updates Tool Output Serialization and Gemini Schema Handling

Vercel has released updates to its AI SDK, including fixes for serializing tool output JSON values and preserving JSON Schema for Google's Gemini models.

By Illumora Editorial

Source · Sep 15, 2026, 4:12 AM · On Illumora · Sep 15, 2026, 4:24 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/workflow-harness@1.0.111 · vercel/ai
Save

Vercel has released updates to its AI SDK, specifically version @ai-sdk/workflow-harness@1.0.111, @ai-sdk/workflow@2.0.32, @ai-sdk/sandbox-vercel@1.0.111, and @ai-sdk/sandbox-just-bash@1.0.111. These releases address issues related to tool output serialization and schema handling for Gemini models.

Key Points

  • The AI SDK now serializes tool output JSON values correctly, as indicated by patch change 6aa7c54 in ai@7.0.101.
  • For @ai-sdk/google@4.0.70, JSON Schema is preserved instead of being converted to OpenAPI schema (patch change 2a32459).
  • A fix ensures that functionResponse.response remains an object for Gemini models, which expect a google.protobuf.Struct (patch change d93e295).
  • Previously, non-object tool results could cause Gemini to close the socket with error 1007.
  • Non-object results are now wrapped under an output key, aligning with Gemini's expected google.protobuf.Struct format.

Context

According to Vercel, these updates were released via the Changesets release GitHub action. The ai@7.0.101 package received a patch to fix tool output JSON serialization. The @ai-sdk/google@4.0.70 package specifically addressed how JSON schemas are handled and how functionResponse.response objects are managed for Gemini models.

Why It Matters

These updates improve the reliability of tool use and function calling within the Vercel AI SDK, particularly for developers integrating with Gemini models. Correct serialization and schema handling prevent unexpected disconnections and ensure data integrity when exchanging information between the SDK and Gemini.

What To Do

  • Developers using the Vercel AI SDK should update to ai@7.0.101 or newer to benefit from the tool output serialization fix.
  • Users of @ai-sdk/google should update to version @ai-sdk/google@4.0.70 to ensure proper Gemini integration.
  • Review the updated documentation for onToolCall and addToolOutput if encountering issues with non-object tool results in Gemini interactions.

Keep Exploring

/atlas/**gemini**-family