Random Errors 429 RESOURCE_EXHAUSTED or Too Many Request

[samuelexferri]

I'm using the library with the gemini-3.1-pro-preview model in the global location in VertexAI mode (not Gemini API or AI Studio). Over the past month, I've been hitting a lot of 429 RESOURCE_EXHAUSTED errors (this was happening with 3.0-pro as well).

I tried enabling retryOptions to handle the load, but I'm still getting Too Many Requests failures. I've already checked my quotas in the Google Cloud Console, and nothing appears to be exceeding 100%. Has anyone else experienced this with the global endpoint?

      this.client = new GoogleGenAI({
        vertexai: this.useVertex,
        project: googleProjectId,
        location: 'global',
        googleAuthOptions: {
          credentials: JSON.parse(googleKeyFile || '{}')
        },
        //httpOptions: { retryOptions: { attempts: 5 } }
      });


      const result = await this.client.models.generateContent({
        model: 'gemini-3.1-pro-preview',
        contents: geminiHistory,
        config: {
          temperature: 1.0,
          systemInstruction: systemPrompt,
          responseMimeType: 'application/json',
          responseSchema: sanitizeJSONSchema(schema)
        }
      });

Environment details

• Programming language: JS/TS
• OS: Ubuntu
• Language runtime version: NodeJS v24
• Package version: 1.42.0

[Varun-S10]

Hi @samuelexferri, I have added you to a private Gist where you can share your Google Cloud project ID, so we can check this issue.

[samuelexferri]

Done!

[Varun-S10]

Hi @samuelexferri, We have passed this issue to our internal team for further investigation. Thank you for your patience.

[dani-mp]

we've also seen this, when in reality we're not even close to the limits

related, we're also seeing tons of these errors that are not legit per our investigations: #1385

[mdyszynski2]

Same issue.

Was using Gemini API with gemini-2.5-flash-image and gemini-3.1-flash-image-preview and suddenly got 429 RESOURCE_EXHAUSTED on 99% of requests for gemini-2.5-flash-image only (gemini-3.1-flash-image-preview continued working normally).

Not even close to any limits on either model.

Switched to VertexAI mode with global location and that seems to have fixed it for now (it's only been a few hours, we'll see).

[mdyszynski2]

Still having the issue today. However, the error now also returns a link: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/provisioned-throughput/error-code-429

It appears that this error can be returned when the model is under heavy load (even if you are under your quota) and there are no resources to process your request. To avoid this you need to purchase provisioned throughput so that your requests are prioritised over pay-as-you-go.