Sign in
← AI PulseSep 15, 2026

Wire · news · Multi-source brief

Vercel AI SDK Updates Tool Output Serialization and Google Gemini Schema Handling

Vercel's AI SDK version 7.0.101 includes a fix for serializing tool output JSON values, while the @ai-sdk/google@4.0.70 package addresses issues with JSON Schema preservation and functionResponse.response object handling for 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/tui@1.0.102 · vercel/ai
Save

Vercel has released updates to its AI SDK, with version 7.0.101 of the ai package and version 4.0.70 of the @ai-sdk/google package. These updates address specific issues related to tool output serialization and interaction with Google's Gemini models.

Key Points

  • The ai@7.0.101 package includes a fix for serializing tool output JSON values.
  • The @ai-sdk/google@4.0.70 package now preserves JSON Schema instead of converting it to OpenAPI schema.
  • A fix in @ai-sdk/google@4.0.70 ensures that functionResponse.response remains an object for Gemini models.
  • Gemini's google.protobuf.Struct field expects an object, and non-object tool results previously caused socket closures.
  • Non-object tool results are now wrapped under an output key as prescribed by the field's documentation.

Context

According to the Vercel AI SDK Changelog, the ai@7.0.101 patch specifically fixes how tool output JSON values are serialized. For the @ai-sdk/google@4.0.70 package, two fixes were implemented: one to preserve JSON Schema directly rather than converting it to OpenAPI schema, and another to ensure that the functionResponse.response field remains an object. This latter fix addresses an issue where Gemini models, which type this field as a google.protobuf.Struct expecting an object, would close the socket if a string, number, array, or null tool result was sent unwrapped.

Why It Matters

These updates enhance the reliability and correctness of tool usage within applications built with the Vercel AI SDK, particularly when interacting with Google's Gemini models. Builders can expect more stable tool call execution and accurate schema handling, reducing unexpected disconnections or data transformation errors.

What To Do

  • Note the update to ai@7.0.101 for improved tool output serialization.
  • If using Google models, update to @ai-sdk/google@4.0.70 to benefit from JSON Schema preservation.
  • Test tool calls with Gemini models after updating to @ai-sdk/google@4.0.70 to confirm stable functionResponse.response handling.
  • Observe how non-object tool results are now wrapped under an output key when interacting with Gemini.