GatewayResponseError: Invalid error response format: Gateway request failed 😵

[RickRyan26]

Description

I suddently started getting this error that me and my ai friends can't solve for the life of us lol.
Would love some thoughts!

250 |   const parseResult = await safeValidateTypes({
251 |     value: response,
252 |     schema: gatewayErrorResponseSchema
253 |   });
254 |   if (!parseResult.success) {
255 |     return new GatewayResponseError({
                     ^
GatewayResponseError: Invalid error response format: Gateway request failed
 statusCode: 200,
       type: "response_error",
   response: {},
 vercel.ai.gateway.error: true,
 vercel.ai.gateway.error.GatewayResponseError: true,

      at createGatewayErrorFromResponse (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:255:16)
      at processTicksAndRejections (native:7:39)

735 |       if (error instanceof Error) {
736 |         if (isAbortError(error) || APICallError3.isInstance(error)) {
737 |           throw error;
738 |         }
739 |       }
740 |       throw new APICallError3({
                  ^
AI_APICallError: Failed to process successful response
      cause: 597 | } from "eventsource-parser/stream";
598 | function parseJsonEventStream({
599 |   stream,
600 |   schema
601 | }) {
602 |   return stream.pipeThrough(new TextDecoderStream()).pipeThrough(new EventSourceParserStream()).pipeThrough(
                ^
TypeError: Invalid state: ReadableStream is locked
 code: "ERR_INVALID_STATE"

      at parseJsonEventStream (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:602:17)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:889:12)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:882:57)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:729:20)
,
        url: "https://ai-gateway.vercel.sh/v1/ai/language-model",
 requestBodyValues: {
  maxOutputTokens: undefined,
  temperature: undefined,
  topP: undefined,
  topK: undefined,
  presencePenalty: undefined,
  frequencyPenalty: undefined,
  stopSequences: undefined,
  seed: undefined,
  tools: [
    [Object ...], [Object ...]
  ],
  toolChoice: [Object ...],
  responseFormat: undefined,
  prompt: [
    [Object ...], [Object ...], [Object ...], [Object ...], [Object ...]
  ],
  providerOptions: undefined,
  headers: undefined,
  includeRawChunks: false,
},
 statusCode: 200,
 responseHeaders: {
  "access-control-allow-headers": "Content-Type, Authorization, ai-gateway-auth-method, ai-gateway-protocol-version, ai-model-id, ai-language-model-id, ai-language-model-specification-version, ai-language-model-streaming, http-referer, x-title",
  "access-control-allow-methods": "GET, POST, PUT, PATCH, DELETE, OPTIONS",
  "access-control-allow-origin": "*",
  "cache-control": "public, max-age=0, must-revalidate",
  date: "Thu, 16 Oct 2025 20:14:31 GMT",
  server: "Vercel",
  "strict-transport-security": "max-age=63072000; includeSubDomains; preload",
  "transfer-encoding": "chunked",
  "x-generation-id": "gen_01K7QAHSYACZ4316MN9X4MVRAG",
  "x-matched-path": "/v1/ai/language-model",
  "x-vercel-cache": "MISS",
  "x-vercel-id": "sfo1::iad1::dr49p-1760645670752-a8eb1fd09c3e",
},
 responseBody: undefined,
 isRetryable: false,
       data: undefined,
 vercel.ai.error: true,
 vercel.ai.error.AI_APICallError: true,

      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:740:13)

597 | } from "eventsource-parser/stream";
598 | function parseJsonEventStream({
599 |   stream,
600 |   schema
601 | }) {
602 |   return stream.pipeThrough(new TextDecoderStream()).pipeThrough(new EventSourceParserStream()).pipeThrough(
                ^
TypeError: Invalid state: ReadableStream is locked
 code: "ERR_INVALID_STATE"

      at parseJsonEventStream (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:602:17)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:889:12)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:882:57)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:729:20)

319 |    */
320 |   static wrap({
321 |     value,
322 |     cause
323 |   }) {
324 |     return _TypeValidationError.isInstance(cause) && cause.value === value ? cause : new _TypeValidationError({ value, cause });
                                                                                           ^
AI_TypeValidationError: Type validation failed: Value: {}.
Error message: [
  {
    "expected": "object",
    "code": "invalid_type",
    "path": [
      "error"
    ],
    "message": "Invalid input: expected object, received undefined"
  }
]
      cause: 25 |     class Definition extends Parent {
26 |     }
27 |     Object.defineProperty(Definition, "name", { value: name });
28 |     function _(def) {
29 |         var _a;
30 |         const inst = params?.Parent ? new Definition() : this;
                                           ^
ZodError: [
  {
    "expected": "object",
    "code": "invalid_type",
    "path": [
      "error"
    ],
    "message": "Invalid input: expected object, received undefined"
  }
]
      at new ZodError (/home/rr26/projects/raicky/node_modules/zod/v4/core/core.js:30:39)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/zod/v4/core/parse.js:53:20)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/zod/v4/core/parse.js:45:49)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:2207:33)
      at validate (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:2206:24)
      at safeValidateTypes (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:531:37)
      at safeValidateTypes (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:522:33)
      at createGatewayErrorFromResponse (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:250:29)
      at createGatewayErrorFromResponse (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:243:46)
      at doStream (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:580:19)
,
      value: {},
 vercel.ai.error: true,
 vercel.ai.error.AI_TypeValidationError: true,

      at wrap (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider/dist/index.mjs:324:86)
      at safeValidateTypes (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:537:35)

25 |     class Definition extends Parent {
26 |     }
27 |     Object.defineProperty(Definition, "name", { value: name });
28 |     function _(def) {
29 |         var _a;
30 |         const inst = params?.Parent ? new Definition() : this;
                                           ^
ZodError: [
  {
    "expected": "object",
    "code": "invalid_type",
    "path": [
      "error"
    ],
    "message": "Invalid input: expected object, received undefined"
  }
]
      at new ZodError (/home/rr26/projects/raicky/node_modules/zod/v4/core/core.js:30:39)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/zod/v4/core/parse.js:53:20)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/zod/v4/core/parse.js:45:49)
      at <anonymous> (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:2207:33)
      at validate (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:2206:24)
      at safeValidateTypes (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:531:37)
      at safeValidateTypes (/home/rr26/projects/raicky/node_modules/@ai-sdk/provider-utils/dist/index.mjs:522:33)
      at createGatewayErrorFromResponse (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:250:29)
      at createGatewayErrorFromResponse (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:243:46)
      at doStream (/home/rr26/projects/raicky/node_modules/@ai-sdk/gateway/dist/index.mjs:580:19)

AI SDK Version

"@ai-sdk/svelte": "3.0.73",
"ai": "5.0.73",

Code of Conduct

• I agree to follow this project's Code of Conduct

[gr2m]

I suddently started getting this error

when did it start? And can you share a minimal example to reproduce it?

[RickRyan26]

@gr2m Thanks for not leaving me hanging.

As I was creating the minimal example for you I found the source of the issue and it seems to be a legitmate Vercel/Sveltekit bug.

Commenting out the following solves the problem... Thoughts?

//src/instrumentation.server.ts

import { registerOTel } from '@vercel/otel'

registerOTel({
	serviceName: 'raicky'
})

[dvaJi]

I had the same issue, and it got fixed after commenting out the instrumentation.ts file.

This is what I had in my code:

import { registerOTel } from "@vercel/otel";
import { LangfuseExporter } from "langfuse-vercel";

export function register() {
  // registerOTel({ serviceName: 'ai-chatbot' });
  if (process.env.LANGFUSE_SECRET_KEY) {
    registerOTel({
      serviceName: "myliai",
      traceExporter: new LangfuseExporter({
        secretKey: process.env.LANGFUSE_SECRET_KEY,
        publicKey: process.env.LANGFUSE_PUBLIC_KEY,
        baseUrl: process.env.LANGFUSE_BASEURL,
      }),
    });
  }
}

[zsiegel92]

I also see this error.

I have agents that execute on the order of ~100 steps each; one out of a few dozen fails with this error.

I think since we're building on the Vercel AI "stack" it would be nice if generateText etc. handled this kind of error if it's indeed due to the AI Gateway (which I think it is).

In case it's helpful, below is some of my stack trace:

GatewayResponseError
Invalid error response format: Gateway request failed



Full error details:

Invalid error response format: Gateway request failed
Stack trace:

GatewayResponseError: Invalid error response format: Gateway request failed
at createGatewayErrorFromResponse (my-repo/node_modules/.pnpm/@ai-sdk+gateway@1.0.15_zod@3.25.76/node_modules/@ai-sdk/gateway/src/errors/create-gateway-error.ts:28:12)
at asGatewayError (my-repo/node_modules/.pnpm/@ai-sdk+gateway@1.0.15_zod@3.25.76/node_modules/@ai-sdk/gateway/src/errors/as-gateway-error.ts:14:12)
at GatewayLanguageModel.doGenerate (my-repo/node_modules/.pnpm/@ai-sdk+gateway@1.0.15_zod@3.25.76/node_modules/@ai-sdk/gateway/src/gateway-language-model.ts:89:13)
at async fn (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/generate-text/generate-text.ts:382:32)
at async <anonymous> (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/telemetry/record-span.ts:18:22)
at async _retryWithExponentialBackoff (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/util/retry-with-exponential-backoff.ts:96:12)
at async fn (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/generate-text/generate-text.ts:338:34)
at async <anonymous> (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/telemetry/record-span.ts:18:22)
at async generateText (my-repo/node_modules/.pnpm/ai@5.0.30_zod@3.25.76/node_modules/ai/src/generate-text/generate-text.ts:274:12)

[abakumov-v]

Hi everyone!
This strange issue happened for me as well today:

[3] [INFO][ai-chat][e59be81f][/api/ai/chat] Request matched route {"pattern":"/api/ai/chat"}
[3] [INFO][ai-chat][e59be81f][/api/ai/chat] Verifying auth token
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat] Auth successful
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] Fetched onboarding data {"hasOnboarding":true,"role":"creator","contentTypesCount":4,"brandVoiceType":"custom"}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] Starting AI chat
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] Loaded conversation history {"messageCount":3,"processedCount":3,"hasToolCalls":false}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 🔧 Tools configuration {"hasTools":true,"toolNames":["format_post_content","format_poll"],"toolCount":2}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 🔑 GATEWAY API KEY CHECK {"hasKey":true,"keyLength":60,"keyPrefix":"vck_1tN3h6MiLd5...","keySuffix":"...qhMn4OnNDb"}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 🤖 MODEL CONFIGURATION {"fullModelString":"openai/gpt-5.1-instant","provider":"openai","modelId":"gpt-5.1-instant"}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] ✅ Gateway model created {"modelConstructorName":"GatewayLanguageModel","hasModel":true}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 💬 MESSAGE FILTERING {"originalCount":3,"filteredCount":3,"filtered":0}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 📋 STREAMTEXT CONFIG SUMMARY {"hasModel":true,"messageCount":4,"maxOutputTokens":2000,"hasTemperature":false,"hasTools":true,"toolCount":2,"hasToolChoice":true,"hasStopWhen":true,"isReasoningModel":true}
[3] [DEBUG][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] Input messages for AI {"tools":["format_post_content","format_poll"],"messageCount":4,"messages":[{"role":"system","content":"\n## CRITICAL: Never Reveal Internal Tools or Processes\n\nNever mention or imply the use of internal tools, functions, or process steps in any user-facing response. \n...(some other system stuff here)"},{"role":"user","content":"give me some ideas"},{"role":"user","content":"try again"},{"role":"user","content":"one more time"}]}
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 🚀 CALLING streamText() NOW...
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] ✅ streamText() returned result object
[3] ✘ [ERROR] Uncaught Error: internal error; reference = 6eu4upfc4kmncbbm8ckb29af
[3]
[3]
[3] [INFO][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] 🔄 Creating UI stream response...
[3] ✘ [ERROR] GatewayResponseError: Invalid error response format: Gateway request failed: internal error; reference = c5jjfu1gjdu54bomjq3puoo5
[3]
[3]       at createGatewayErrorFromResponse (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/errors/create-gateway-error.ts:39:12)
[3]       at async GatewayLanguageModel.doStream (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/gateway-language-model.ts:162:13)
[3]       at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1162:27)
[3]       ... 4 lines matching cause stack trace ...
[3]       at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22) {
[3]     statusCode: 500,
[3]     cause: Error: internal error; reference = c5jjfu1gjdu54bomjq3puoo5
[3]         at async postToApi (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/provider-utils/src/post-to-api.ts:98:22)
[3]         at async GatewayLanguageModel.doStream (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/gateway-language-model.ts:102:52)
[3]         at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1162:27)
[3]         at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22)
[3]         at async _retryWithExponentialBackoff (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/util/retry-with-exponential-backoff.ts:96:12)
[3]         at async streamStep (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1113:15)
[3]         at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1535:9)
[3]         at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22) {
[3]       remote: true
[3]     },
[3]     type: 'response_error',
[3]     response: {},
[3]     validationError: _TypeValidationError [AI_TypeValidationError]: Type validation failed: Value: {}.
[3]     Error message: [
[3]       {
[3]         "expected": "object",
[3]         "code": "invalid_type",
[3]         "path": [
[3]           "error"
[3]         ],
[3]         "message": "Invalid input: expected object, received undefined"
[3]       }
[3]     ]
[3]         at _TypeValidationError.wrap (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/provider/src/errors/type-validation-error.ts:49:9)
[3]         at safeValidateTypes (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/provider-utils/src/validate-types.ts:71:34)
[3]         at async createGatewayErrorFromResponse (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/errors/create-gateway-error.ts:33:23)
[3]         at async GatewayLanguageModel.doStream (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/gateway-language-model.ts:162:13)
[3]         at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1162:27)
[3]         at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22)
[3]         at async _retryWithExponentialBackoff (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/util/retry-with-exponential-backoff.ts:96:12)
[3]         at async streamStep (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1113:15)
[3]         at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1535:9)
[3]         at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22) {
[3]       cause: ZodError
[3]           at new ZodError (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/zod/v4/core/core.js:30:39)
[3]           at safeParseAsync (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/zod/v4/core/parse.js:53:20)
[3]           at Object.validate (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/provider-utils/src/zod-schema.ts:67:33)
[3]           at safeValidateTypes (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/provider-utils/src/validate-types.ts:63:36)
[3]           at createGatewayErrorFromResponse (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/errors/create-gateway-error.ts:33:29)
[3]           at asGatewayError (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/errors/as-gateway-error.ts:23:10)
[3]           at GatewayLanguageModel.doStream (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/@ai-sdk/gateway/src/gateway-language-model.ts:162:19)
[3]           at async fn (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/generate-text/stream-text.ts:1162:27)
[3]           at null.<anonymous> (async file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/.wrangler/tmp/dev-YWXNUD/index.js:23993:22)
[3]           at async _retryWithExponentialBackoff (file:///C:/Users/av-19/Documents/projects/github/abakumov-v/posts-scheduler-bolt/workers/ai-chat/node_modules/ai/src/util/retry-with-exponential-backoff.ts:96:12) {
[3]         message: [Getter]
[3]       },
[3]       value: {},
[3]       Symbol(vercel.ai.error): true,
[3]       Symbol(vercel.ai.error.AI_TypeValidationError): true
[3]     },
[3]     Symbol(vercel.ai.gateway.error): true,
[3]     Symbol(vercel.ai.gateway.error.GatewayResponseError): true
[3]   }
[3]
[3]
[3] [wrangler:info] POST /api/ai/chat 200 OK (2962ms)
[3] ✘ [ERROR] [ERROR][ai-chat][e59be81f][user:c112684a][/api/ai/chat][conv:af5bb596] Error in onFinish callback {"error":"No output generated. Check the stream for errors."}

and it happens only for local Cloudflare worker!
And If I deploy it to cloud Cloudflare Worker, it works fine.
I started to face with it only today, and spend a lot of time to figure out why it happens - still have no idea.

I even tried to reset my branch to a state when it definetely worked locally (13 Jan), and the error is still the same.

Maybe, someone can give me some clue?

[philiparpin]

We started seeing this completely randomly without any changes today using the gateway. Nothing we could tell that would explain it even remotely. Switching from using gateway to using anthropic directly resolved it. No reported outages either. This alone may be the reason we never try gateway again. Worked fine completely locally but in a deployed environment it failed consistently. Disabled all tracing with no luck whatsoever.

[philiparpin]

Look at the timing:

DNS: Works fine -- resolves to 64.239.123.129 and 64.239.109.129 (Vercel's edge IPs via Cloudflare). No IPv6 issues.
TLS: ECONNRESET in 36ms
36ms is the smoking gun. That's not a timeout or a network issue -- that's an active rejection. The TCP connection succeeds (SYN/SYN-ACK/ACK), then Render sends the TLS ClientHello, and Vercel's edge immediately sends back a TCP RST. The connection is being deliberately killed.

Vercel's edge (Cloudflare) is looking at the source IP from Render and resetting the connection before TLS negotiation completes. A real network problem would timeout (seconds), not reset in 36ms. A TLS version mismatch would send a TLS alert, not a raw TCP RST.

So to directly answer your question -- the gateway itself is fine. It never even sees your requests. Cloudflare, sitting in front of Vercel's edge, is dropping connections from Render's IP range at the TCP layer. Whether that's intentional policy or collateral damage from abuse on shared Render IPs, only Vercel/Cloudflare can tell you.

Going direct to api.anthropic.com works because Anthropic's infrastructure doesn't have the same IP-level blocking.

[lgrammel]

Identify issue type

{
  "evidence": [
    "Reporter gets `GatewayResponseError: Invalid error response format: Gateway request failed` with statusCode 200, followed by `AI_APICallError: Failed to process successful response` and `TypeError: Invalid state: ReadableStream is locked`.",
    "Reporter later says commenting out `registerOTel` from `@vercel/otel` resolves the problem, and another user confirms the same with `instrumentation.ts`.",
    "Multiple commenters report the same GatewayResponseError occurring intermittently or in specific environments, including Cloudflare Worker local/dev and deployed Render environments."
  ],
  "rationale": "The issue reports unexpected runtime failures from AI SDK/Gateway APIs during normal `streamText`/`generateText` usage, including invalid error parsing and stream locking errors. Even though the root cause may involve OTel or infrastructure, the issue is framed around broken current behavior and thrown errors.",
  "type": "bug",
  "confidence": "medium"
}