@ai-sdk/google 4.0.82 → 4.0.84

package/CHANGELOG.md

@@ -1,5 +1,31 @@
 # @ai-sdk/google
 
+## 4.0.84
+
+### Patch Changes
+
+- d0290cf: fix(google): preserve realtime user utterances with independent IDs and cumulative transcripts, honor text-free transcription completion markers, and advance response IDs after interruption
+
+## 4.0.83
+
+### Patch Changes
+
+- 32cf2f6: fix(google): send the default Gemini Live `thinkingLevel` when `thinkingConfig` sets neither `thinkingLevel` nor `thinkingBudget`, and stop overwriting a raw `generationConfig.thinkingConfig`
+- 525efc5: feat(provider): advertise image model file and mask input support
+
+  Use confirmed model IDs for capability declarations so unrecognized model names
+  remain unknown. Include Together AI FLUX.2 Pro and Flex single-image editing, and
+  allow asynchronous capability lookups and middleware overrides to resolve to
+  unknown.
+
+  Advertise QuiverAI Arrow 2 and Arrow 2 Telos file-input support, and mark Together
+  AI Gemini image inputs unsupported by the current single-image request mapping.
+
+- Updated dependencies [e3605f6]
+- Updated dependencies [525efc5]
+  - @ai-sdk/provider-utils@5.0.50
+  - @ai-sdk/provider@4.0.19
+
 ## 4.0.82
 
 ### Patch Changes

package/dist/index.js

@@ -8,7 +8,7 @@ import {
 import { Experimental_EvaluationLanguageModel as EvaluationLanguageModel } from "@ai-sdk/provider-utils/experimental-evaluation";
 
 // src/version.ts
-var VERSION = true ? "4.0.82" : "0.0.0-test";
+var VERSION = true ? "4.0.84" : "0.0.0-test";
 
 // src/google-embedding-model.ts
 import {
@@ -3912,6 +3912,11 @@ import {
   WORKFLOW_SERIALIZE as WORKFLOW_SERIALIZE3,
   WORKFLOW_DESERIALIZE as WORKFLOW_DESERIALIZE3
 } from "@ai-sdk/provider-utils";
+var googleImageModelsWithFileInputSupport = /* @__PURE__ */ new Set([
+  "gemini-2.5-flash-image",
+  "gemini-3-pro-image-preview",
+  "gemini-3.1-flash-image-preview"
+]);
 var GoogleImageModel = class _GoogleImageModel {
   constructor(modelId, settings, config) {
     this.modelId = modelId;
@@ -3934,6 +3939,12 @@ var GoogleImageModel = class _GoogleImageModel {
     }
     return 1;
   }
+  get supportsFileInputs() {
+    return googleImageModelsWithFileInputSupport.has(this.modelId) ? true : void 0;
+  }
+  get supportsMaskInputs() {
+    return this.supportsFileInputs === true ? false : void 0;
+  }
   get provider() {
     return this.config.provider;
   }
@@ -8080,6 +8091,10 @@ import { isRecord, safeParseJSON } from "@ai-sdk/provider-utils";
 var GoogleRealtimeEventMapper = class {
   constructor() {
     this.turnCounter = 0;
+    this.inputTranscriptionCounter = 0;
+    this.inputTranscriptionBuffer = "";
+    this.inputTranscriptionBoundary = "none";
+    this.preserveInputTranscriptionAcrossTurnComplete = false;
     this.hasAudio = false;
     this.hasText = false;
     this.hasTranscript = false;
@@ -8093,12 +8108,11 @@ var GoogleRealtimeEventMapper = class {
     return `google-item-${this.turnCounter}`;
   }
   /**
-   * Rolls over to the next turn lazily, only once new model content actually
-   * arrives. `turnComplete` merely marks the current turn closed; the counter
-   * is not advanced until the next response begins. This keeps a transcript
-   * that arrives shortly after `turnComplete` attached to the turn it belongs
-   * to, since Google delivers transcription independently with no guaranteed
-   * ordering relative to `turnComplete`.
+   * Rolls over to the next turn lazily, once model content for the next turn
+   * arrives. `turnComplete` merely marks the current turn closed.
+   * This keeps an output transcript that arrives shortly after `turnComplete`
+   * attached to the turn it belongs to, since Google delivers transcription
+   * independently with no guaranteed ordering relative to `turnComplete`.
    */
   beginTurnIfClosed() {
     if (!this.turnClosed) return;
@@ -8107,6 +8121,48 @@ var GoogleRealtimeEventMapper = class {
     this.hasText = false;
     this.hasTranscript = false;
     this.turnClosed = false;
+    this.preserveInputTranscriptionAcrossTurnComplete = false;
+  }
+  // Google streams input transcription as a sequence of non-accumulating
+  // fragments for a single user utterance (e.g. "The quick brown fox",
+  // " jumps over the", " lazy dog."). Concatenate consecutive fragments into
+  // one running transcript under a stable synthetic id so the realtime reducer
+  // (which overwrites the message for a given id) surfaces them as a single
+  // coherent user message rather than one message per fragment. Google's
+  // `finished` signal is the authoritative utterance boundary. For older
+  // payloads without that signal, response boundaries remain a fallback.
+  //
+  // Input transcriptions can arrive independently of response turn events, so
+  // their synthetic IDs use a sequence independent of `turnCounter`.
+  accumulateInputTranscription({
+    text,
+    finished
+  }) {
+    if (!text) {
+      if (finished === true) {
+        this.finishInputTranscription();
+      }
+      return void 0;
+    }
+    const hasFinishedSignal = finished != null;
+    if (this.inputTranscriptionBuffer !== "" && (this.inputTranscriptionBoundary === "interrupted" || this.inputTranscriptionBoundary === "turn-complete" && !hasFinishedSignal)) {
+      this.finishInputTranscription();
+    }
+    this.inputTranscriptionBoundary = "none";
+    this.inputTranscriptionBuffer += text;
+    const result = {
+      itemId: `google-input-${this.inputTranscriptionCounter}`,
+      transcript: this.inputTranscriptionBuffer
+    };
+    if (finished === true) {
+      this.finishInputTranscription();
+    }
+    return result;
+  }
+  finishInputTranscription() {
+    if (this.inputTranscriptionBuffer === "") return;
+    this.inputTranscriptionCounter++;
+    this.inputTranscriptionBuffer = "";
   }
   parseServerEvent(raw) {
     const data = raw;
@@ -8163,19 +8219,26 @@ var GoogleRealtimeEventMapper = class {
     if (data.serverContent != null) {
       return this.parseServerContent(data.serverContent, raw);
     }
-    if (data.inputTranscription?.text != null) {
-      return {
-        type: "input-transcription-completed",
-        itemId: `google-input-${this.turnCounter}`,
-        transcript: data.inputTranscription.text,
-        raw
-      };
+    if (data.inputTranscription != null) {
+      const transcription = this.accumulateInputTranscription(
+        data.inputTranscription
+      );
+      if (transcription != null) {
+        return {
+          type: "input-transcription-completed",
+          ...transcription,
+          raw
+        };
+      }
     }
     return { type: "custom", rawType: String(Object.keys(data)[0]), raw };
   }
   parseServerContent(serverContent, raw) {
     const events = [];
     if (serverContent.interrupted) {
+      this.inputTranscriptionBoundary = "interrupted";
+      this.preserveInputTranscriptionAcrossTurnComplete = true;
+      this.turnClosed = true;
       events.push({
         type: "speech-started",
         raw
@@ -8216,13 +8279,17 @@ var GoogleRealtimeEventMapper = class {
         raw
       });
     }
-    if (serverContent.inputTranscription?.text) {
-      events.push({
-        type: "input-transcription-completed",
-        itemId: `google-input-${this.turnCounter}`,
-        transcript: serverContent.inputTranscription.text,
-        raw
-      });
+    if (serverContent.inputTranscription != null) {
+      const transcription = this.accumulateInputTranscription(
+        serverContent.inputTranscription
+      );
+      if (transcription != null) {
+        events.push({
+          type: "input-transcription-completed",
+          ...transcription,
+          raw
+        });
+      }
     }
     if (serverContent.generationComplete) {
       events.push({
@@ -8277,6 +8344,13 @@ var GoogleRealtimeEventMapper = class {
         raw
       });
       this.turnClosed = true;
+      if (this.preserveInputTranscriptionAcrossTurnComplete) {
+        this.preserveInputTranscriptionAcrossTurnComplete = false;
+      } else {
+        if (this.inputTranscriptionBoundary !== "interrupted") {
+          this.inputTranscriptionBoundary = "turn-complete";
+        }
+      }
     }
     if (events.length === 0) {
       return { type: "custom", rawType: "serverContent", raw };
@@ -8405,7 +8479,9 @@ function buildGoogleSessionConfig(config, modelId) {
   if (config?.outputAudioTranscription != null) {
     setup.outputAudioTranscription = {};
   }
-  const thinkingConfig = googleOptions?.thinkingConfig ?? (isThinkingLiveModel(modelId) ? { thinkingLevel: "low" } : void 0);
+  const rawGenerationConfig = restProviderOptions.generationConfig;
+  const explicitThinkingConfig = googleOptions?.thinkingConfig ?? (isRecord(rawGenerationConfig) && isRecord(rawGenerationConfig.thinkingConfig) ? rawGenerationConfig.thinkingConfig : void 0);
+  const thinkingConfig = isThinkingLiveModel(modelId) && explicitThinkingConfig?.thinkingLevel == null && explicitThinkingConfig?.thinkingBudget == null ? { ...explicitThinkingConfig, thinkingLevel: "low" } : googleOptions?.thinkingConfig;
   const applyThinkingConfig = () => {
     if (thinkingConfig == null) return;
     const target = isRecord(setup.generationConfig) ? setup.generationConfig : generationConfig;