models/imagen-3.0-fast-generate-001 is not found for API version v1beta, or is not supported for predict

[bnicholl]

When attempting to use imagen-3.0-fast-generate-001 I get the error

ClientError: 404 NOT_FOUND. {'error': {'code': 404, 'message': 'models/imagen-3.0-fast-generate-001 is not found for API version v1beta, or is not supported for predict. Call ListModels to see the list of available models and their supported methods.', 'status': 'NOT_FOUND'}}

Steps to reproduce

from google import genai
from google.genai import types
from PIL import Image
from io import BytesIO
import base64

client = genai.Client(api_key='###')


response = client.models.generate_images(
    #model='imagen-3.0-generate-002',
    model = 'imagen-3.0-fast-generate-001',
    prompt= "give me an image backdrop for a childrens book. The premise is santa clause hires new helpers. ",
    config=types.GenerateImagesConfig(
        number_of_images= 1,
    )
)

Note that if I change the model id to imagen-3.0-generate-002, it works fine. According to this documentation

https://cloud.google.com/vertex-ai/generative-ai/docs/models#imagen-models

imagen-3.0-fast-generate-001 should be generally available, so not sure why it isn't working.

[matthew29tang]

The documentation you linked is for Vertex AI. You can use Vertex AI by specifying vertexai=True in the client initialization with a project and location (right now you are using an API key for Gemini Developer API)

Here is the documentation for Gemini Developer API Image generation models: https://ai.google.dev/gemini-api/docs/models

This page correctly lists that imagen-3.0-generate-002 is the only model supported.

[bnicholl]

Is there any reason why this isn't available on the Gemini API? Trying t deploy anything on AWS using Vertex API is more work than I'm willing to do. It'd be really nice to just have an API key I can use In a lambda function(which is easily accomplished with gemini API), as opposed to try and figure out how to get google IAM roles working inside of AWS.

[matthew29tang]

I haven't heard of any plans to support an Imagen 3 fast endpoint on Gemini API, sorry.

By the way, Vertex also supports API keys: https://cloud.google.com/vertex-ai/generative-ai/docs/start/express-mode/overview

[bnicholl]

Thanks for the response. Nice to know vertex also supports API keys. Correct me if I'm wrong, but after looking at the documentation you posted above, if looks like none of the Imagen models are supported on Vertex AI for express mode, which is the only way to use an API key with Vertex. I'm really curious as to why Google has all of these different API's and API rules for their products. It makes navigating google dev products much more tedious and confusing.

[matthew29tang]

Sorry to hear that, hopefully this unified GenAI SDK is a step in the right direction so you can at least easily switch between backends (Vertex, Gemini Developer API) with the switch of a boolean.

I did hear about Vertex having plans to support API keys outside of express mode (and support all models) which was landing soon, let me check on that.

[bnicholl]

Thanks, I really appreciate your attentiveness to these issues. I'll be looking forward to hearing about the new Vertex functionality when it's available.

[matthew29tang]

The new API key support docs can be found here (make sure to click yes I have an account, otherwise it will show gcp express instructions): https://cloud.google.com/vertex-ai/generative-ai/docs/start/api-keys?usertype=existinguser

Hopefully this solves your issue!

[delbyte]

hey, it's not working with imagen-3.0-generate-002 anymore either! @matthew29tang

[Google Imagen] Request: {
  model: 'imagen-3.0-generate-002',
  modelId: 'imagen-3',
  promptLength: 9,
  config: { numberOfImages: 1, aspectRatio: '1:1' }
}
[Google Imagen] Error: {
  message: '{"error":{"code":404,"message":"models/imagen-3.0-generate-002 is not found for API version v1beta, or is not supported for predict. Call ListModels to see the list of available models and their supported methods.","status":"NOT_FOUND"}}',
  details: tv [ApiError]: {"error":{"code":404,"message":"models/imagen-3.0-generate-002 is not found for API version v1beta, or is not supported for predict. Call ListModels to see the list of available models and their supported methods.","status":"NOT_FOUND"}}

...

Error: tv [ApiError]: {"error":{"code":404,"message":"models/imagen-3.0-generate-002 is not found for API version v1beta, or is not supported for predict. Call ListModels to see the list of available models and their supported methods.","status":"NOT_FOUND"}}
...

could we access these models again without going through the hassle of integrating Vertex AI?

[matthew29tang]

For gemini developer API, imagen-3.0-generate-002 has been deprecated and turned down per https://ai.google.dev/gemini-api/docs/changelog#11-10-2025

Please use Imagen 4 instead (imagen-4.0-generate-001)

[delbyte]

Thank you! I'm sorry I should have looked that up beforehand 😓