{"error":{"code":429,"message":"Resource has been exhausted (e.g. check quota).","status":"RESOURCE_EXHAUSTED"}}

[oojr]

I am on a paid tier and I am getting resource exhausted trying to analyze a single image, text generations work with the same api

Environment details

• Programming language: Javascript
• OS: Mac M1
• Language runtime version: NodeJS 24.15.0
• Package version: 2.16.0

Steps to reproduce

const {
  GoogleGenAI,
  createUserContent,
  createPartFromUri,
} = require("@google/genai");

const GEMINI_API_TOKEN =
  process.env.GEMINI_API_KEY;

const ai = new GoogleGenAI({ apiKey: GEMINI_API_TOKEN });

async function runTest(imageUrl, mimeType = "image/jpeg") {
  const parts = [
    "Describe this receipt in one sentence.",
    createPartFromUri(imageUrl, mimeType),
  ];

  try {
    const result = await ai.models.generateContent({
      model: "gemini-3.5-flash-lite",
      contents: createUserContent(parts),
    });

    console.log(result.text);
    return result.text;
  } catch (error) {
    console.error("[auditReceiptImage Error]", error);
    throw error;
  }
}

[oojr]

it seems I have to upload the file to google servers first, this is a new change

const {
  GoogleGenAI,
  Type,
  createUserContent,
  createPartFromUri, // <-- Import createPartFromUri
} = require("@google/genai");

const GEMINI_API_TOKEN =
  process.env.GEMINI_API_KEY

const ai = new GoogleGenAI({ apiKey: GEMINI_API_TOKEN });

async function runTest(imageUrl, mimeType = "image/jpeg") {
  try {
    // 1. Fetch remote image as Blob
    const response = await fetch(imageUrl);
    if (!response.ok) {
      throw new Error(`Failed to fetch image: ${response.statusText}`);
    }
    const blob = await response.blob();

    // 2. Upload file to Gemini Files API
    const uploadedFile = await ai.files.upload({
      file: blob,
      mimeType: mimeType,
      config: {
        displayName: "receipt.jpg",
      },
    });

    console.log(`Uploaded file URI: ${uploadedFile.uri}`);

    // 3. Convert uploadedFile.uri into a valid Part object
    const filePart = createPartFromUri(
      uploadedFile.uri,
      uploadedFile.mimeType || mimeType,
    );

    const result = await ai.models.generateContent({
      model: "gemini-3.5-flash",
      contents: createUserContent([
        "Describe this receipt in one sentence.",
        filePart,
      ]),
    });

    console.log("Model Output:", result.text);
    return result.text;
  } catch (error) {
    console.error("[auditReceiptImage Error]", error);
    throw error;
  }
}

runTest(
  "https://pub-ce48c8741312498b952de4673252c236.r2.dev/uploads/47f50a54-0f57-400e-bbb6-ccc557fa24e7-B4B9C135-83ED-4506-8A71-971229880AD2.jpg",
);

[Varun-S10]

Hi @oojr, Please follow the below instructions to troubleshoot:

Go to the Google Cloud Console and navigate to APIs & Services.
Select Generative Language API (or Vertex AI API).
Under the Quotas & System Limits tab, check the Current Usage percentage.
If usage reaches 100%, you have reached your quota limit, which results in the 429 (Resource Exhausted) error.

If you think there is any discrepancy or need further assistance, please let us know.

[oojr]

@Varun-S10 my quota is not reach but when I look at the logs I see URL_REJECTED-REJECTED_CLIENT_CAPABILITY, remote urls are now being rejected? When I use the file upload api with google before doing an image analysis it works but is more cpu intensive as I need to make a blob on the server side, I guess something changed on the api side

[DavidMaza]

I also have this exact same problem with the "429 Resource has been exhausted (e.g. check quota)" error. It started happening out of the blue in the past week or two.

I am below the spend cap configured in Google AI Studio, the prepaid balance on my account is positive and I only make 1 or 2 queries to Gemini API a day. So I don't think I'm hitting any API usage limits.

I'm also including an image in my prompts to Gemini. I've tried it with a small image, about 200 pixels x 200 pixels large. The image is a PNG image with an AWS S3 presigned URL.

I raised a ticket with Google but it didn't get much traction unfortunately after a few emails back and forth. The ticket number was 74201059.

Removing the image from the "generateContent()" request got me past the error and I received a successful response. So for some reason I believe the Gemini API is suddenly rejecting externally hosted images and throwing a "429 Resource has been exhausted (e.g. check quota)" error as a result of it.

I can also see the URL_REJECTED-REJECTED_CLIENT_CAPABILITY error code when I look at my Gemini Logs in the Google AI Studio.

Below is some sample code:

import {Injectable} from '@angular/core';
import {from, map, Observable} from 'rxjs';
import {createPartFromUri, GoogleGenAI} from '@google/genai';

@Injectable({
  providedIn: 'root'
})
export class GeminiService {
  client = new GoogleGenAI({
    apiKey: 'zzzz'
  });

  analyse(
    presignedImageUrl: string,
    imageMimeType: string
  ): Observable<string | undefined> {
    const image = createPartFromUri(presignedImageUrl, imageMimeType);

    return from(
      this.client.models.generateContent({
        model: 'gemini-3.1-pro',
        contents: ['The quick brown fox jumped over the lazy dog', image]
      })
    ).pipe(
      map(result => {
        return result.text;
      })
    );
  }
}

Another discussion where the error was recently reported: https://discuss.ai.google.dev/t/error-429-url-rejected-rejected-client-capability/177893/2

[urko27]

Same issue, when sending the file inline as base64 there is no problem. But sending a presigned url it returns 429.

[oojr]

@urko27 @DavidMaza you need to upload the file using client.file.uploads first before using the api now

[oojr]

@DavidMaza angular example, I vibe coded this so test it

import { Injectable } from '@angular/core';
import { from, map, Observable, switchMap } from 'rxjs';
import { createPartFromUri, createUserContent, GoogleGenAI } from '@google/genai';

@Injectable({
  providedIn: 'root'
})
export class GeminiService {
  client = new GoogleGenAI({
    apiKey: 'zzzz'
  });

  analyse(
    presignedImageUrl: string,
    imageMimeType: string,
    prompt: string = 'Describe this receipt in one sentence.'
  ): Observable<string | undefined> {
    return from(this.uploadAndAnalyse(presignedImageUrl, imageMimeType, prompt));
  }

  private async uploadAndAnalyse(
    presignedImageUrl: string,
    imageMimeType: string,
    prompt: string
  ): Promise<string | undefined> {
    // 1. Fetch remote image as Blob
    const response = await fetch(presignedImageUrl);
    if (!response.ok) {
      throw new Error(`Failed to fetch image: ${response.statusText}`);
    }
    const blob = await response.blob();

    // 2. Upload file to Gemini Files API
    const uploadedFile = await this.client.files.upload({
      file: blob,
      mimeType: imageMimeType,
      config: {
        displayName: 'receipt.jpg',
      },
    });

    // 3. Convert uploadedFile.uri into a valid Part object
    const filePart = createPartFromUri(
      uploadedFile.uri!,
      uploadedFile.mimeType || imageMimeType,
    );

    // 4. Generate content
    const result = await this.client.models.generateContent({
      model: 'gemini-3.1-pro',
      contents: createUserContent([prompt, filePart]),
    });

    return result.text;
  }
}

[urko27]

It still shows up in the docs

https://ai.google.dev/gemini-api/docs/file-input-methods#external-urls

[Varun-S10]

Thank you for bringing this to our attention. We truly appreciate you flagging this issue, we will file a bug internally.