chore: remove logprobs (#5896)

Author: lgrammel

.changeset/itchy-deers-jog.md

@@ -0,0 +1,5 @@
+---
+'@ai-sdk/provider': major
+---
+
+chore: remove logprobs

examples/ai-core/src/generate-text/openai-logprobs.ts

@@ -36,7 +36,6 @@ async function main() {
 
       case 'finish': {
         console.log('Finish reason:', part.finishReason);
-        console.log('Logprobs:', part.logprobs);
         console.log('Usage:', part.usage);
         break;
       }

examples/ai-core/src/stream-object/openai-fullstream.ts

@@ -92,13 +92,6 @@ exports[`streamObject > output = "object" > result.fullStream > should send full
   },
   {
     "finishReason": "stop",
-    "logprobs": [
-      {
-        "logprob": 1,
-        "token": "-",
-        "topLogprobs": [],
-      },
-    ],
     "response": {
       "id": "id-0",
       "modelId": "mock-model-id",

examples/ai-core/src/stream-text/azure-fullstream-logprobs.ts

@@ -3,7 +3,6 @@ import {
   FinishReason,
   LanguageModelRequestMetadata,
   LanguageModelResponseMetadata,
-  LogProbs,
   ProviderMetadata,
 } from '../types';
 import { LanguageModelUsage } from '../types/usage';
@@ -47,14 +46,6 @@ Response body (available only for providers that use HTTP requests).
     body?: unknown;
   };
 
-  /**
- Logprobs for the completion.
-`undefined` if the mode does not support logprobs or if was not enabled.
-
-@deprecated Will become a provider extension in the future.
-     */
-  readonly logprobs: LogProbs | undefined;
-
   /**
 Additional provider-specific metadata. They are passed through
 from the provider to the AI SDK and enable provider-specific

examples/ai-core/src/stream-text/openai-fullstream-logprobs.ts

@@ -23,7 +23,6 @@ import {
   CallWarning,
   FinishReason,
   LanguageModel,
-  LogProbs,
   ProviderMetadata,
 } from '../types';
 import { LanguageModelRequestMetadata } from '../types/language-model-request-metadata';
@@ -264,7 +263,6 @@ export async function generateObject<SCHEMA, RESULT>({
       let warnings: CallWarning[] | undefined;
       let response: LanguageModelResponseMetadata;
       let request: LanguageModelRequestMetadata;
-      let logprobs: LogProbs | undefined;
       let resultProviderMetadata: ProviderMetadata | undefined;
 
       const standardizedPrompt = standardizePrompt({
@@ -377,7 +375,6 @@ export async function generateObject<SCHEMA, RESULT>({
       finishReason = generateResult.finishReason;
       usage = generateResult.usage;
       warnings = generateResult.warnings;
-      logprobs = generateResult.logprobs;
       resultProviderMetadata = generateResult.providerMetadata;
       request = generateResult.request ?? {};
       response = generateResult.responseData;
@@ -468,7 +465,6 @@ export async function generateObject<SCHEMA, RESULT>({
         warnings,
         request,
         response,
-        logprobs,
         providerMetadata: resultProviderMetadata,
       });
     },
@@ -480,7 +476,6 @@ class DefaultGenerateObjectResult<T> implements GenerateObjectResult<T> {
   readonly finishReason: GenerateObjectResult<T>['finishReason'];
   readonly usage: GenerateObjectResult<T>['usage'];
   readonly warnings: GenerateObjectResult<T>['warnings'];
-  readonly logprobs: GenerateObjectResult<T>['logprobs'];
   readonly providerMetadata: GenerateObjectResult<T>['providerMetadata'];
   readonly response: GenerateObjectResult<T>['response'];
   readonly request: GenerateObjectResult<T>['request'];
@@ -490,7 +485,6 @@ class DefaultGenerateObjectResult<T> implements GenerateObjectResult<T> {
     finishReason: GenerateObjectResult<T>['finishReason'];
     usage: GenerateObjectResult<T>['usage'];
     warnings: GenerateObjectResult<T>['warnings'];
-    logprobs: GenerateObjectResult<T>['logprobs'];
     providerMetadata: GenerateObjectResult<T>['providerMetadata'];
     response: GenerateObjectResult<T>['response'];
     request: GenerateObjectResult<T>['request'];
@@ -502,7 +496,6 @@ class DefaultGenerateObjectResult<T> implements GenerateObjectResult<T> {
     this.providerMetadata = options.providerMetadata;
     this.response = options.response;
     this.request = options.request;
-    this.logprobs = options.logprobs;
   }
 
   toJsonResponse(init?: ResponseInit): Response {

packages/ai/core/generate-object/__snapshots__/stream-object.test.ts.snap

@@ -4,7 +4,6 @@ import {
   FinishReason,
   LanguageModelRequestMetadata,
   LanguageModelResponseMetadata,
-  LogProbs,
   ProviderMetadata,
 } from '../types';
 import { LanguageModelUsage } from '../types/usage';
@@ -108,7 +107,6 @@ export type ObjectStreamPart<PARTIAL> =
   | {
       type: 'finish';
       finishReason: FinishReason;
-      logprobs?: LogProbs;
       usage: LanguageModelUsage;
       response: LanguageModelResponseMetadata;
       providerMetadata?: ProviderMetadata;

packages/ai/core/generate-object/generate-object-result.ts

@@ -223,7 +223,6 @@ describe('streamObject', () => {
                   type: 'finish',
                   finishReason: 'stop',
                   usage: { inputTokens: 2, outputTokens: 10 },
-                  logprobs: [{ token: '-', logprob: 1, topLogprobs: [] }],
                 },
               ]),
             }),
@@ -440,7 +439,6 @@ describe('streamObject', () => {
                 {
                   type: 'finish',
                   finishReason: 'stop',
-                  logprobs: undefined,
                   usage: { inputTokens: 3, outputTokens: 10 },
                 },
               ]),
@@ -482,7 +480,6 @@ describe('streamObject', () => {
                 {
                   type: 'finish',
                   finishReason: 'stop',
-                  logprobs: undefined,
                   usage: { inputTokens: 3, outputTokens: 10 },
                 },
               ]),