diff --git a/.size-limit.js b/.size-limit.js
index 38a83445d021..5bf0e534fbda 100644
--- a/.size-limit.js
+++ b/.size-limit.js
@@ -326,7 +326,7 @@ module.exports = [
     import: createImport('init'),
     ignore: [...builtinModules, ...nodePrefixedBuiltinModules],
     gzip: true,
-    limit: '175 KB',
+    limit: '176 KB',
   },
   {
     name: '@sentry/node - without tracing',
diff --git a/dev-packages/e2e-tests/test-applications/nextjs-15/tests/ai-test.test.ts b/dev-packages/e2e-tests/test-applications/nextjs-15/tests/ai-test.test.ts
index 9fd05f83c5f9..a53f8986512a 100644
--- a/dev-packages/e2e-tests/test-applications/nextjs-15/tests/ai-test.test.ts
+++ b/dev-packages/e2e-tests/test-applications/nextjs-15/tests/ai-test.test.ts
@@ -34,14 +34,14 @@ test('should create AI spans with correct attributes', async ({ page }) => {
   expect(firstPipelineSpan?.data?.['vercel.ai.model.id']).toBe('mock-model-id');
   expect(firstPipelineSpan?.data?.['vercel.ai.model.provider']).toBe('mock-provider');
   expect(firstPipelineSpan?.data?.['vercel.ai.prompt']).toContain('Where is the first span?');
-  expect(firstPipelineSpan?.data?.['gen_ai.response.text']).toBe('First span here!');
+  expect(firstPipelineSpan?.data?.['gen_ai.output.messages']).toContain('First span here!');
   expect(firstPipelineSpan?.data?.['gen_ai.usage.input_tokens']).toBe(10);
   expect(firstPipelineSpan?.data?.['gen_ai.usage.output_tokens']).toBe(20); */
 
   // Second AI call - explicitly enabled telemetry
   const secondPipelineSpan = aiPipelineSpans[0];
   expect(secondPipelineSpan?.data?.['vercel.ai.prompt']).toContain('Where is the second span?');
-  expect(secondPipelineSpan?.data?.['gen_ai.response.text']).toContain('Second span here!');
+  expect(secondPipelineSpan?.data?.['gen_ai.output.messages']).toContain('Second span here!');
 
   // Third AI call - with tool calls
   /*  const thirdPipelineSpan = aiPipelineSpans[2];
diff --git a/dev-packages/e2e-tests/test-applications/nextjs-16/tests/ai-test.test.ts b/dev-packages/e2e-tests/test-applications/nextjs-16/tests/ai-test.test.ts
index f7dc95e7d00d..5c519cb89a03 100644
--- a/dev-packages/e2e-tests/test-applications/nextjs-16/tests/ai-test.test.ts
+++ b/dev-packages/e2e-tests/test-applications/nextjs-16/tests/ai-test.test.ts
@@ -34,14 +34,14 @@ test('should create AI spans with correct attributes', async ({ page }) => {
   expect(firstPipelineSpan?.data?.['vercel.ai.model.id']).toBe('mock-model-id');
   expect(firstPipelineSpan?.data?.['vercel.ai.model.provider']).toBe('mock-provider');
   expect(firstPipelineSpan?.data?.['vercel.ai.prompt']).toContain('Where is the first span?');
-  expect(firstPipelineSpan?.data?.['gen_ai.response.text']).toBe('First span here!');
+  expect(firstPipelineSpan?.data?.['gen_ai.output.messages']).toContain('First span here!');
   expect(firstPipelineSpan?.data?.['gen_ai.usage.input_tokens']).toBe(10);
   expect(firstPipelineSpan?.data?.['gen_ai.usage.output_tokens']).toBe(20); */
 
   // Second AI call - explicitly enabled telemetry
   const secondPipelineSpan = aiPipelineSpans[0];
   expect(secondPipelineSpan?.data?.['vercel.ai.prompt']).toContain('Where is the second span?');
-  expect(secondPipelineSpan?.data?.['gen_ai.response.text']).toContain('Second span here!');
+  expect(secondPipelineSpan?.data?.['gen_ai.output.messages']).toContain('Second span here!');
 
   // Third AI call - with tool calls
   /*  const thirdPipelineSpan = aiPipelineSpans[2];
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/scenario.mjs b/dev-packages/node-integration-tests/suites/tracing/vercelai/scenario.mjs
index 9bfdd4a9793a..b6abe6fdf673 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/scenario.mjs
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/scenario.mjs
@@ -51,6 +51,7 @@ async function run() {
       }),
       tools: {
         getWeather: {
+          description: 'Get the current weather for a location',
           parameters: z.object({ location: z.string() }),
           execute: async args => {
             return `Weather in ${args.location}: Sunny, 72°F`;
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/test.ts b/dev-packages/node-integration-tests/suites/tracing/vercelai/test.ts
index 2919815b8f0d..809ba2308622 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/test.ts
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/test.ts
@@ -5,16 +5,16 @@ import {
   GEN_AI_INPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE,
   GEN_AI_OPERATION_NAME_ATTRIBUTE,
+  GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE,
   GEN_AI_REQUEST_MODEL_ATTRIBUTE,
   GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE,
   GEN_AI_RESPONSE_ID_ATTRIBUTE,
   GEN_AI_RESPONSE_MODEL_ATTRIBUTE,
-  GEN_AI_RESPONSE_TEXT_ATTRIBUTE,
-  GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE,
   GEN_AI_SYSTEM_ATTRIBUTE,
   GEN_AI_SYSTEM_INSTRUCTIONS_ATTRIBUTE,
   GEN_AI_TOOL_CALL_ID_ATTRIBUTE,
+  GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE,
   GEN_AI_TOOL_INPUT_ATTRIBUTE,
   GEN_AI_TOOL_NAME_ATTRIBUTE,
   GEN_AI_TOOL_OUTPUT_ATTRIBUTE,
@@ -91,9 +91,10 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the second span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
           [GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 30,
@@ -119,11 +120,12 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['stop'],
           [GEN_AI_RESPONSE_ID_ATTRIBUTE]: expect.any(String),
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_SYSTEM_ATTRIBUTE]: 'mock-provider',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -201,6 +203,7 @@ describe('Vercel AI integration', () => {
         status: 'ok',
       }),
       // Seventh span - tool call execution span
+      // Note: gen_ai.tool.description is NOT present when sendDefaultPii: false because ai.prompt.tools is not recorded
       expect.objectContaining({
         data: {
           [GEN_AI_TOOL_CALL_ID_ATTRIBUTE]: 'call-1',
@@ -220,7 +223,7 @@ describe('Vercel AI integration', () => {
   };
 
   const EXPECTED_AVAILABLE_TOOLS_JSON =
-    '[{"type":"function","name":"getWeather","parameters":{"type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false,"$schema":"http://json-schema.org/draft-07/schema#"}}]';
+    '[{"type":"function","name":"getWeather","description":"Get the current weather for a location","parameters":{"type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false,"$schema":"http://json-schema.org/draft-07/schema#"}}]';
 
   const EXPECTED_TRANSACTION_DEFAULT_PII_TRUE = {
     transaction: 'main',
@@ -230,9 +233,10 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the first span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
           [GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 30,
@@ -264,11 +268,12 @@ describe('Vercel AI integration', () => {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]:
             '[{"role":"user","content":[{"type":"text","text":"Where is the first span?"}]}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['stop'],
           [GEN_AI_RESPONSE_ID_ATTRIBUTE]: expect.any(String),
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           [GEN_AI_SYSTEM_ATTRIBUTE]: 'mock-provider',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -302,9 +307,10 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the second span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
           [GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 30,
@@ -335,11 +341,12 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['stop'],
           [GEN_AI_RESPONSE_ID_ATTRIBUTE]: expect.any(String),
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_SYSTEM_ATTRIBUTE]: 'mock-provider',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -373,10 +380,10 @@ describe('Vercel AI integration', () => {
         data: {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"What is the weather in San Francisco?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Tool call completed!"},{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{ \\"location\\": \\"San Francisco\\" }"}],"finish_reason":"tool_call"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'Tool call completed!',
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 15,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 25,
           [GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 40,
@@ -408,12 +415,12 @@ describe('Vercel AI integration', () => {
           [GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE]: EXPECTED_AVAILABLE_TOOLS_JSON,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Tool call completed!"},{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{ \\"location\\": \\"San Francisco\\" }"}],"finish_reason":"tool_call"}]',
           [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['tool-calls'],
           [GEN_AI_RESPONSE_ID_ATTRIBUTE]: expect.any(String),
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'Tool call completed!',
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           [GEN_AI_SYSTEM_ATTRIBUTE]: 'mock-provider',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 15,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 25,
@@ -447,6 +454,7 @@ describe('Vercel AI integration', () => {
       expect.objectContaining({
         data: {
           [GEN_AI_TOOL_CALL_ID_ATTRIBUTE]: 'call-1',
+          [GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE]: 'Get the current weather for a location',
           [GEN_AI_TOOL_INPUT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_TOOL_NAME_ATTRIBUTE]: 'getWeather',
           [GEN_AI_TOOL_OUTPUT_ATTRIBUTE]: expect.any(String),
@@ -809,7 +817,6 @@ describe('Vercel AI integration', () => {
                   data: expect.objectContaining({
                     [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 3,
                     [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.stringMatching(/^\[.*"(?:text|content)":"C+".*\]$/),
-                    [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'Response to truncated messages',
                   }),
                 }),
                 // Second call: Last message is small and kept intact
@@ -819,7 +826,6 @@ describe('Vercel AI integration', () => {
                     [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.stringContaining(
                       'This is a small message that fits within the limit',
                     ),
-                    [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'Response to small message',
                   }),
                 }),
               ]),
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/scenario.mjs b/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/scenario.mjs
index 9ef1b8000741..2c83234064ae 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/scenario.mjs
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/scenario.mjs
@@ -47,6 +47,7 @@ async function run() {
       }),
       tools: {
         getWeather: tool({
+          description: 'Get the current weather for a location',
           inputSchema: z.object({ location: z.string() }),
           execute: async ({ location }) => `Weather in ${location}: Sunny, 72°F`,
         }),
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/test.ts b/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/test.ts
index 7d981a878363..a84b80e9abc5 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/test.ts
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/v5/test.ts
@@ -5,15 +5,15 @@ import {
   GEN_AI_INPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE,
   GEN_AI_OPERATION_NAME_ATTRIBUTE,
+  GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE,
   GEN_AI_REQUEST_MODEL_ATTRIBUTE,
   GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE,
   GEN_AI_RESPONSE_ID_ATTRIBUTE,
   GEN_AI_RESPONSE_MODEL_ATTRIBUTE,
-  GEN_AI_RESPONSE_TEXT_ATTRIBUTE,
-  GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE,
   GEN_AI_SYSTEM_ATTRIBUTE,
   GEN_AI_TOOL_CALL_ID_ATTRIBUTE,
+  GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE,
   GEN_AI_TOOL_INPUT_ATTRIBUTE,
   GEN_AI_TOOL_NAME_ATTRIBUTE,
   GEN_AI_TOOL_OUTPUT_ATTRIBUTE,
@@ -93,7 +93,8 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.pipeline.name': 'generateText',
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the second span?"}]',
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
@@ -127,7 +128,8 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.model': 'mock-model-id',
           'vercel.ai.response.id': expect.any(String),
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.timestamp': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
@@ -196,6 +198,7 @@ describe('Vercel AI integration (V5)', () => {
         status: 'ok',
       }),
       // Seventh span - tool call execution span
+      // Note: gen_ai.tool.description is NOT present when sendDefaultPii: false because ai.prompt.tools is not recorded
       expect.objectContaining({
         data: {
           'vercel.ai.operationId': 'ai.toolCall',
@@ -215,7 +218,7 @@ describe('Vercel AI integration (V5)', () => {
   };
 
   const EXPECTED_AVAILABLE_TOOLS_JSON =
-    '[{"type":"function","name":"getWeather","inputSchema":{"$schema":"http://json-schema.org/draft-07/schema#","type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false}}]';
+    '[{"type":"function","name":"getWeather","description":"Get the current weather for a location","inputSchema":{"$schema":"http://json-schema.org/draft-07/schema#","type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false}}]';
 
   const EXPECTED_TRANSACTION_DEFAULT_PII_TRUE = {
     transaction: 'main',
@@ -230,8 +233,9 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the first span?"}]',
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the first span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -257,10 +261,11 @@ describe('Vercel AI integration (V5)', () => {
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]:
             '[{"role":"user","content":[{"type":"text","text":"Where is the first span?"}]}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.id': expect.any(String),
           'vercel.ai.response.model': 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           'vercel.ai.response.timestamp': expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
@@ -290,8 +295,9 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the second span?"}]',
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the second span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -323,7 +329,8 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.model': 'mock-model-id',
           'vercel.ai.response.id': expect.any(String),
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.timestamp': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
@@ -349,8 +356,9 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"What is the weather in San Francisco?"}]',
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"What is the weather in San Francisco?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{\\"location\\":\\"San Francisco\\"}"}],"finish_reason":"tool_call"}]',
           'vercel.ai.response.finishReason': 'tool-calls',
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -375,14 +383,14 @@ describe('Vercel AI integration (V5)', () => {
           'vercel.ai.pipeline.name': 'generateText.doGenerate',
           [GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE]: 1,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{\\"location\\":\\"San Francisco\\"}"}],"finish_reason":"tool_call"}]',
           'vercel.ai.prompt.toolChoice': expect.any(String),
           [GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE]: EXPECTED_AVAILABLE_TOOLS_JSON,
           'vercel.ai.response.finishReason': 'tool-calls',
           'vercel.ai.response.id': expect.any(String),
           'vercel.ai.response.model': 'mock-model-id',
-          // 'gen_ai.response.text': 'Tool call completed!', // TODO: look into why this is not being set
           'vercel.ai.response.timestamp': expect.any(String),
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['tool-calls'],
@@ -406,6 +414,7 @@ describe('Vercel AI integration (V5)', () => {
         data: expect.objectContaining({
           'vercel.ai.operationId': 'ai.toolCall',
           [GEN_AI_TOOL_CALL_ID_ATTRIBUTE]: 'call-1',
+          [GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE]: 'Get the current weather for a location',
           [GEN_AI_TOOL_NAME_ATTRIBUTE]: 'getWeather',
           [GEN_AI_TOOL_INPUT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_TOOL_OUTPUT_ATTRIBUTE]: expect.any(String),
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/scenario.mjs b/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/scenario.mjs
index 66233d1dabe5..ee2dc802cd9c 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/scenario.mjs
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/scenario.mjs
@@ -62,6 +62,7 @@ async function run() {
       }),
       tools: {
         getWeather: tool({
+          description: 'Get the current weather for a location',
           inputSchema: z.object({ location: z.string() }),
           execute: async ({ location }) => `Weather in ${location}: Sunny, 72°F`,
         }),
diff --git a/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/test.ts b/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/test.ts
index 2a213f39410d..39ee00254373 100644
--- a/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/test.ts
+++ b/dev-packages/node-integration-tests/suites/tracing/vercelai/v6/test.ts
@@ -4,15 +4,15 @@ import { afterAll, describe, expect } from 'vitest';
 import {
   GEN_AI_INPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_OPERATION_NAME_ATTRIBUTE,
+  GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE,
   GEN_AI_REQUEST_MODEL_ATTRIBUTE,
   GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE,
   GEN_AI_RESPONSE_ID_ATTRIBUTE,
   GEN_AI_RESPONSE_MODEL_ATTRIBUTE,
-  GEN_AI_RESPONSE_TEXT_ATTRIBUTE,
-  GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE,
   GEN_AI_SYSTEM_ATTRIBUTE,
   GEN_AI_TOOL_CALL_ID_ATTRIBUTE,
+  GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE,
   GEN_AI_TOOL_INPUT_ATTRIBUTE,
   GEN_AI_TOOL_NAME_ATTRIBUTE,
   GEN_AI_TOOL_OUTPUT_ATTRIBUTE,
@@ -95,10 +95,11 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the second span?"}]',
           'vercel.ai.request.headers.user-agent': expect.any(String),
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the second span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -129,9 +130,10 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.model': 'mock-model-id',
           'vercel.ai.response.id': expect.any(String),
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           'vercel.ai.response.timestamp': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['stop'],
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -199,6 +201,7 @@ describe('Vercel AI integration (V6)', () => {
         status: 'ok',
       }),
       // Seventh span - tool call execution span
+      // Note: gen_ai.tool.description is NOT present when sendDefaultPii: false because ai.prompt.tools is not recorded
       expect.objectContaining({
         data: expect.objectContaining({
           'vercel.ai.operationId': 'ai.toolCall',
@@ -218,7 +221,7 @@ describe('Vercel AI integration (V6)', () => {
   };
 
   const EXPECTED_AVAILABLE_TOOLS_JSON =
-    '[{"type":"function","name":"getWeather","inputSchema":{"$schema":"http://json-schema.org/draft-07/schema#","type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false}}]';
+    '[{"type":"function","name":"getWeather","description":"Get the current weather for a location","inputSchema":{"$schema":"http://json-schema.org/draft-07/schema#","type":"object","properties":{"location":{"type":"string"}},"required":["location"],"additionalProperties":false}}]';
 
   const EXPECTED_TRANSACTION_DEFAULT_PII_TRUE = {
     transaction: 'main',
@@ -233,8 +236,9 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the first span?"}]',
           'vercel.ai.request.headers.user-agent': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the first span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -260,10 +264,11 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.request.headers.user-agent': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]:
             '[{"role":"user","content":[{"type":"text","text":"Where is the first span?"}]}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"First span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.id': expect.any(String),
           'vercel.ai.response.model': 'mock-model-id',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: 'First span here!',
           'vercel.ai.response.timestamp': expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
@@ -293,8 +298,9 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"Where is the second span?"}]',
           'vercel.ai.request.headers.user-agent': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"Where is the second span?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           'vercel.ai.response.finishReason': 'stop',
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -327,9 +333,10 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.response.finishReason': 'stop',
           'vercel.ai.response.model': 'mock-model-id',
           'vercel.ai.response.id': expect.any(String),
-          [GEN_AI_RESPONSE_TEXT_ATTRIBUTE]: expect.any(String),
           'vercel.ai.response.timestamp': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"text","content":"Second span here!"}],"finish_reason":"stop"}]',
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['stop'],
           [GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
           [GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 20,
@@ -352,8 +359,9 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.prompt': '[{"role":"user","content":"What is the weather in San Francisco?"}]',
           'vercel.ai.request.headers.user-agent': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: '[{"role":"user","content":"What is the weather in San Francisco?"}]',
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{\\"location\\":\\"San Francisco\\"}"}],"finish_reason":"tool_call"}]',
           'vercel.ai.response.finishReason': 'tool-calls',
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'mock-model-id',
@@ -378,14 +386,14 @@ describe('Vercel AI integration (V6)', () => {
           'vercel.ai.pipeline.name': 'generateText.doGenerate',
           'vercel.ai.request.headers.user-agent': expect.any(String),
           [GEN_AI_INPUT_MESSAGES_ATTRIBUTE]: expect.any(String),
+          [GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE]:
+            '[{"role":"assistant","parts":[{"type":"tool_call","id":"call-1","name":"getWeather","arguments":"{\\"location\\":\\"San Francisco\\"}"}],"finish_reason":"tool_call"}]',
           'vercel.ai.prompt.toolChoice': expect.any(String),
           [GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE]: EXPECTED_AVAILABLE_TOOLS_JSON,
           'vercel.ai.response.finishReason': 'tool-calls',
           'vercel.ai.response.id': expect.any(String),
           'vercel.ai.response.model': 'mock-model-id',
-          // 'gen_ai.response.text': 'Tool call completed!', // TODO: look into why this is not being set
           'vercel.ai.response.timestamp': expect.any(String),
-          [GEN_AI_RESPONSE_TOOL_CALLS_ATTRIBUTE]: expect.any(String),
           'vercel.ai.settings.maxRetries': 2,
           'vercel.ai.streaming': false,
           [GEN_AI_RESPONSE_FINISH_REASONS_ATTRIBUTE]: ['tool-calls'],
@@ -409,6 +417,7 @@ describe('Vercel AI integration (V6)', () => {
         data: expect.objectContaining({
           'vercel.ai.operationId': 'ai.toolCall',
           [GEN_AI_TOOL_CALL_ID_ATTRIBUTE]: 'call-1',
+          [GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE]: 'Get the current weather for a location',
           [GEN_AI_TOOL_NAME_ATTRIBUTE]: 'getWeather',
           [GEN_AI_TOOL_INPUT_ATTRIBUTE]: expect.any(String),
           [GEN_AI_TOOL_OUTPUT_ATTRIBUTE]: expect.any(String),
diff --git a/packages/core/src/tracing/ai/gen-ai-attributes.ts b/packages/core/src/tracing/ai/gen-ai-attributes.ts
index dc88e6315852..4f8d4b0161c2 100644
--- a/packages/core/src/tracing/ai/gen-ai-attributes.ts
+++ b/packages/core/src/tracing/ai/gen-ai-attributes.ts
@@ -126,6 +126,14 @@ export const GEN_AI_INPUT_MESSAGES_ORIGINAL_LENGTH_ATTRIBUTE = 'sentry.sdk_meta.
  */
 export const GEN_AI_INPUT_MESSAGES_ATTRIBUTE = 'gen_ai.input.messages';
 
+/**
+ * The model's response messages including text and tool calls
+ * Only recorded when recordOutputs is enabled
+ * Format: stringified array of message objects with role, parts, and finish_reason
+ * @see https://opentelemetry.io/docs/specs/semconv/registry/attributes/gen-ai/#gen-ai-output-messages
+ */
+export const GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE = 'gen_ai.output.messages';
+
 /**
  * The system instructions extracted from system messages
  * Only recorded when recordInputs is enabled
@@ -269,6 +277,12 @@ export const GEN_AI_TOOL_INPUT_ATTRIBUTE = 'gen_ai.tool.input';
  */
 export const GEN_AI_TOOL_OUTPUT_ATTRIBUTE = 'gen_ai.tool.output';
 
+/**
+ * The description of the tool being used
+ * @see https://opentelemetry.io/docs/specs/semconv/registry/attributes/gen-ai/#gen-ai-tool-description
+ */
+export const GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE = 'gen_ai.tool.description';
+
 // =============================================================================
 // OPENAI-SPECIFIC ATTRIBUTES
 // =============================================================================
diff --git a/packages/core/src/tracing/ai/mediaStripping.ts b/packages/core/src/tracing/ai/mediaStripping.ts
index f4870cd5a9de..cb8e5d7b959e 100644
--- a/packages/core/src/tracing/ai/mediaStripping.ts
+++ b/packages/core/src/tracing/ai/mediaStripping.ts
@@ -47,6 +47,8 @@ export function isContentMedia(part: unknown): part is ContentMedia {
     hasInputAudio(part) ||
     hasFileData(part) ||
     hasMediaTypeData(part) ||
+    hasVercelFileData(part) ||
+    hasVercelImageData(part) ||
     hasBlobOrBase64Type(part) ||
     hasB64Json(part) ||
     hasImageGenerationResult(part) ||
@@ -113,6 +115,41 @@ function hasMediaTypeData(part: NonNullable<unknown>): part is { media_type: str
   return 'media_type' in part && typeof part.media_type === 'string' && 'data' in part;
 }
 
+/**
+ * Check for Vercel AI SDK file format: { type: "file", mediaType: "...", data: "..." }
+ * Only matches base64/binary data, not HTTP/HTTPS URLs (which should be preserved).
+ */
+function hasVercelFileData(part: NonNullable<unknown>): part is { type: 'file'; mediaType: string; data: string } {
+  return (
+    'type' in part &&
+    part.type === 'file' &&
+    'mediaType' in part &&
+    typeof part.mediaType === 'string' &&
+    'data' in part &&
+    typeof part.data === 'string' &&
+    // Only strip base64/binary data, not HTTP/HTTPS URLs which should be preserved as references
+    !part.data.startsWith('http://') &&
+    !part.data.startsWith('https://')
+  );
+}
+
+/**
+ * Check for Vercel AI SDK image format: { type: "image", image: "base64...", mimeType?: "..." }
+ * Only matches base64/data URIs, not HTTP/HTTPS URLs (which should be preserved).
+ * Note: mimeType is optional in Vercel AI SDK image parts.
+ */
+function hasVercelImageData(part: NonNullable<unknown>): part is { type: 'image'; image: string; mimeType?: string } {
+  return (
+    'type' in part &&
+    part.type === 'image' &&
+    'image' in part &&
+    typeof part.image === 'string' &&
+    // Only strip base64/data URIs, not HTTP/HTTPS URLs which should be preserved as references
+    !part.image.startsWith('http://') &&
+    !part.image.startsWith('https://')
+  );
+}
+
 function hasBlobOrBase64Type(part: NonNullable<unknown>): part is { type: 'blob' | 'base64'; content: string } {
   return 'type' in part && (part.type === 'blob' || part.type === 'base64');
 }
@@ -131,7 +168,7 @@ function hasDataUri(part: NonNullable<unknown>): part is { uri: string } {
 
 const REMOVED_STRING = '[Blob substitute]';
 
-const MEDIA_FIELDS = ['image_url', 'data', 'content', 'b64_json', 'result', 'uri'] as const;
+const MEDIA_FIELDS = ['image_url', 'data', 'content', 'b64_json', 'result', 'uri', 'image'] as const;
 
 /**
  * Replace inline binary data in a single media content part with a placeholder.
diff --git a/packages/core/src/tracing/ai/messageTruncation.ts b/packages/core/src/tracing/ai/messageTruncation.ts
index 499d25ee6e47..aade37b84474 100644
--- a/packages/core/src/tracing/ai/messageTruncation.ts
+++ b/packages/core/src/tracing/ai/messageTruncation.ts
@@ -229,11 +229,53 @@ function truncatePartsMessage(message: PartsMessage, maxBytes: number): unknown[
   }
 }
 
+/**
+ * Truncate a message with `content: [...]` array format (Vercel AI SDK, OpenAI multimodal).
+ * Content arrays contain parts like `{ type: "text", text: "..." }`.
+ *
+ * @param message - Message with content array property
+ * @param maxBytes - Maximum byte limit
+ * @returns Array with truncated message, or empty array if it doesn't fit
+ */
+function truncateContentArrayMessage(message: ContentArrayMessage, maxBytes: number): unknown[] {
+  const { content } = message;
+
+  // Find the first text part to truncate
+  const textPartIndex = content.findIndex(
+    part => part && typeof part === 'object' && 'type' in part && part.type === 'text' && 'text' in part,
+  );
+
+  if (textPartIndex === -1) {
+    // No text part found, cannot truncate safely
+    return [];
+  }
+
+  const textPart = content[textPartIndex] as { type: string; text: string };
+
+  // Calculate overhead (message structure with empty text)
+  const emptyContent = content.map((part, i) => (i === textPartIndex ? { ...textPart, text: '' } : part));
+  const emptyMessage = { ...message, content: emptyContent };
+  const overhead = jsonBytes(emptyMessage);
+  const availableForText = maxBytes - overhead;
+
+  if (availableForText <= 0) {
+    return [];
+  }
+
+  const truncatedText = truncateTextByBytes(textPart.text, availableForText);
+  const truncatedContent = content.map((part, i) =>
+    i === textPartIndex ? { ...textPart, text: truncatedText } : part,
+  );
+
+  return [{ ...message, content: truncatedContent }];
+}
+
 /**
  * Truncate a single message to fit within maxBytes.
  *
- * Supports two message formats:
+ * Supports three message formats:
  * - OpenAI/Anthropic: `{ ..., content: string }`
+ * - Vercel AI/OpenAI multimodal: `{ ..., content: Array<{type, text?, ...}> }`
  * - Google GenAI: `{ ..., parts: Array<string | {text: string} | non-text> }`
  *
  * @param message - The message to truncate
@@ -257,6 +299,10 @@ function truncateSingleMessage(message: unknown, maxBytes: number): unknown[] {
     return truncateContentMessage(message, maxBytes);
   }
 
+  if (isContentArrayMessage(message)) {
+    return truncateContentArrayMessage(message, maxBytes);
+  }
+
   if (isPartsMessage(message)) {
     return truncatePartsMessage(message, maxBytes);
   }
diff --git a/packages/core/src/tracing/vercel-ai/index.ts b/packages/core/src/tracing/vercel-ai/index.ts
index 919c06eb12d6..8b62079295af 100644
--- a/packages/core/src/tracing/vercel-ai/index.ts
+++ b/packages/core/src/tracing/vercel-ai/index.ts
@@ -7,9 +7,12 @@ import { spanToJSON } from '../../utils/spanUtils';
 import {
   GEN_AI_INPUT_MESSAGES_ATTRIBUTE,
   GEN_AI_OPERATION_NAME_ATTRIBUTE,
+  GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE,
+  GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE,
   GEN_AI_REQUEST_MODEL_ATTRIBUTE,
   GEN_AI_RESPONSE_MODEL_ATTRIBUTE,
   GEN_AI_TOOL_CALL_ID_ATTRIBUTE,
+  GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE,
   GEN_AI_TOOL_INPUT_ATTRIBUTE,
   GEN_AI_TOOL_NAME_ATTRIBUTE,
   GEN_AI_TOOL_OUTPUT_ATTRIBUTE,
@@ -42,6 +45,7 @@ import {
   AI_OPERATION_ID_ATTRIBUTE,
   AI_PROMPT_MESSAGES_ATTRIBUTE,
   AI_PROMPT_TOOLS_ATTRIBUTE,
+  AI_RESPONSE_FINISH_REASON_ATTRIBUTE,
   AI_RESPONSE_OBJECT_ATTRIBUTE,
   AI_RESPONSE_PROVIDER_METADATA_ATTRIBUTE,
   AI_RESPONSE_TEXT_ATTRIBUTE,
@@ -124,13 +128,21 @@ function vercelAiEventProcessor(event: Event): Event {
       accumulateTokensForParent(span, tokenAccumulator);
     }
 
-    // Second pass: apply accumulated token data to parent spans
+    // Second pass: apply tool descriptions and accumulated tokens
     for (const span of event.spans) {
-      if (span.op !== 'gen_ai.invoke_agent') {
-        continue;
+      if (span.op === 'gen_ai.execute_tool') {
+        const toolName = span.data[GEN_AI_TOOL_NAME_ATTRIBUTE];
+        if (typeof toolName === 'string') {
+          const description = findToolDescription(event.spans, toolName);
+          if (description) {
+            span.data[GEN_AI_TOOL_DESCRIPTION_ATTRIBUTE] = description;
+          }
+        }
       }
 
-      applyAccumulatedTokens(span, tokenAccumulator);
+      if (span.op === 'gen_ai.invoke_agent') {
+        applyAccumulatedTokens(span, tokenAccumulator);
+      }
     }
 
     // Also apply to root when it is the invoke_agent pipeline
@@ -142,6 +154,140 @@ function vercelAiEventProcessor(event: Event): Event {
 
   return event;
 }
+
+/**
+ * Finds a tool description by scanning spans for gen_ai.request.available_tools
+ * (already processed from ai.prompt.tools in the first pass).
+ */
+function findToolDescription(spans: SpanJSON[], toolName: string): string | undefined {
+  for (const span of spans) {
+    const availableTools = span.data[GEN_AI_REQUEST_AVAILABLE_TOOLS_ATTRIBUTE];
+    if (typeof availableTools !== 'string') {
+      continue;
+    }
+    try {
+      const tools = JSON.parse(availableTools) as Array<{ name?: string; description?: string }>;
+      const tool = tools.find(t => t.name === toolName);
+      if (tool?.description) {
+        return tool.description;
+      }
+    } catch {
+      // ignore
+    }
+  }
+  return undefined;
+}
+
+/**
+ * Tool call structure from Vercel AI SDK
+ * Note: V5/V6 use 'input' for arguments, V4 and earlier use 'args'
+ */
+interface VercelToolCall {
+  toolCallId: string;
+  toolName: string;
+  input?: Record<string, unknown> | string; // V5/V6
+  args?: string; // V4 and earlier
+}
+
+/**
+ * Normalize finish reason to match OpenTelemetry semantic conventions.
+ * Valid values: "stop", "length", "content_filter", "tool_call", "error"
+ *
+ * Vercel AI SDK uses "tool-calls" (plural, with hyphen) which we map to "tool_call".
+ */
+function normalizeFinishReason(finishReason: unknown): string {
+  if (typeof finishReason !== 'string') {
+    return 'stop';
+  }
+
+  // Map Vercel AI SDK finish reasons to OpenTelemetry semantic convention values
+  switch (finishReason) {
+    case 'tool-calls':
+      return 'tool_call';
+    case 'stop':
+    case 'length':
+    case 'content_filter':
+    case 'error':
+      return finishReason;
+    default:
+      // For unknown values, return as-is (schema allows arbitrary strings)
+      return finishReason;
+  }
+}
+
+/**
+ * Build gen_ai.output.messages from ai.response.text and/or ai.response.toolCalls
+ *
+ * Format follows OpenTelemetry semantic conventions:
+ * [{"role": "assistant", "parts": [...], "finish_reason": "stop"}]
+ *
+ * Parts can be:
+ * - {"type": "text", "content": "..."}
+ * - {"type": "tool_call", "id": "...", "name": "...", "arguments": "..."}
+ */
+function buildOutputMessages(attributes: Record<string, unknown>): void {
+  const responseText = attributes[AI_RESPONSE_TEXT_ATTRIBUTE];
+  const responseToolCalls = attributes[AI_RESPONSE_TOOL_CALLS_ATTRIBUTE];
+  const finishReason = attributes[AI_RESPONSE_FINISH_REASON_ATTRIBUTE];
+
+  // Skip if neither text nor tool calls are present
+  if (responseText == null && responseToolCalls == null) {
+    return;
+  }
+
+  const parts: Array<Record<string, unknown>> = [];
+
+  // Add text part if present
+  if (typeof responseText === 'string' && responseText.length > 0) {
+    parts.push({
+      type: 'text',
+      content: responseText,
+    });
+  }
+
+  // Add tool call parts if present
+  if (responseToolCalls != null) {
+    try {
+      // Tool calls can be a string (JSON) or already parsed array
+      const toolCalls: VercelToolCall[] =
+        typeof responseToolCalls === 'string' ? JSON.parse(responseToolCalls) : responseToolCalls;
+
+      if (Array.isArray(toolCalls)) {
+        for (const toolCall of toolCalls) {
+          // V5/V6 use 'input', V4 and earlier use 'args'
+          const args = toolCall.input ?? toolCall.args;
+          parts.push({
+            type: 'tool_call',
+            id: toolCall.toolCallId,
+            name: toolCall.toolName,
+            arguments: typeof args === 'string' ? args : JSON.stringify(args),
+          });
+        }
+      }
+    } catch {
+      // Ignore parsing errors
+    }
+  }
+
+  // Only set output messages and delete source attributes if we have parts
+  // This ensures we don't lose telemetry data if parsing fails
+  if (parts.length > 0) {
+    const outputMessage = {
+      role: 'assistant',
+      parts,
+      finish_reason: normalizeFinishReason(finishReason),
+    };
+
+    attributes[GEN_AI_OUTPUT_MESSAGES_ATTRIBUTE] = JSON.stringify([outputMessage]);
+
+    // Remove the source attributes since they're now captured in gen_ai.output.messages
+    // eslint-disable-next-line @typescript-eslint/no-dynamic-delete
+    delete attributes[AI_RESPONSE_TEXT_ATTRIBUTE];
+    // eslint-disable-next-line @typescript-eslint/no-dynamic-delete
+    delete attributes[AI_RESPONSE_TOOL_CALLS_ATTRIBUTE];
+  }
+}
+
 /**
  * Post-process spans emitted by the Vercel AI SDK.
  */
@@ -196,8 +342,11 @@ function processEndedVercelAiSpan(span: SpanJSON): void {
     delete attributes[OPERATION_NAME_ATTRIBUTE];
   }
   renameAttributeKey(attributes, AI_PROMPT_MESSAGES_ATTRIBUTE, GEN_AI_INPUT_MESSAGES_ATTRIBUTE);
-  renameAttributeKey(attributes, AI_RESPONSE_TEXT_ATTRIBUTE, 'gen_ai.response.text');
-  renameAttributeKey(attributes, AI_RESPONSE_TOOL_CALLS_ATTRIBUTE, 'gen_ai.response.tool_calls');
+
+  // Build gen_ai.output.messages from response text and/or tool calls
+  // Note: buildOutputMessages also removes the source attributes when output is successfully generated
+  buildOutputMessages(attributes);
+
   renameAttributeKey(attributes, AI_RESPONSE_OBJECT_ATTRIBUTE, 'gen_ai.response.object');
   renameAttributeKey(attributes, AI_PROMPT_TOOLS_ATTRIBUTE, 'gen_ai.request.available_tools');