OpenAI Image Analyze gpt model

Nelson_Silva_Filho · May 29, 2024, 2:39pm

Is there a way to choose the gpt model on OpenAI Image Analyze built-in node ?

n8n version: 1.42.1
Database: SQLite):
n8n EXECUTIONS_PROCESS setting: own
Running n8n via: Dcoker
Operating system: Linux

n8n · May 29, 2024, 2:39pm

It looks like your topic is missing some important information. Could you provide the following if applicable.

n8n version:
Database (default: SQLite):
n8n EXECUTIONS_PROCESS setting (default: own, main):
Running n8n via (Docker, npm, n8n cloud, desktop app):
Operating system:

ria · May 31, 2024, 11:51am

Hi @Nelson_Silva_Filho

Thanks for posting here and welcome to the community!

So, for the built-in OpenAI Image Analyze node we were using the GPT-4-Vision because it was the only option that supported it at the time.

github.com

n8n-io/n8n/blob/master/packages/@n8n/nodes-langchain/nodes/vendors/OpenAi/actions/image/analyze.operation.ts

import type {
	INodeProperties,
	IExecuteFunctions,
	INodeExecutionData,
	IDataObject,
} from 'n8n-workflow';
import { updateDisplayOptions, NodeOperationError } from 'n8n-workflow';
import { apiRequest } from '../../transport';

const properties: INodeProperties[] = [
	{
		displayName: 'Text Input',
		name: 'text',
		type: 'string',
		placeholder: "e.g. What's in this image?",
		default: "What's in this image?",
		typeOptions: {
			rows: 2,
		},
	},

This file has been truncated. show original

Would be worth it to update to the new GPT-4o though. Let me raise an internal issue for this.

For the moment, I think you can use the LLM Chain together with the OpenAI Model Sub-Node if you want to specify the model for image analysis. You will have to do some prompt engineering though. Or use the HTTP node to make a direct call to the OpenAI API.

Let us know how you get on!

Nelson_Silva_Filho · June 6, 2024, 11:53am

Thank you!

jb8n · June 7, 2024, 7:02am

Thanks Ria.

Adding to this, it would indeed make sense to offer models for the vision node, or at least upgrade it to gpt4o

Here’s what I just got from openai:

As part of our continuous upgrade process, we'll be [deprecating], our latest flagship model, as a replacement for these models.

Compared to gpt-4-vision-preview, gpt-4o has:

* 2x lower pricing
* 2x the speed
* 5x higher rate limits
* state of the art visual understanding, with an eval score of 69.1% vs. gpt-4-vision-preview’s 63.1%

jan · June 12, 2024, 2:06pm

New version [email protected] got released which includes the GitHub PR 9660.

Fabio_Ferrero · June 20, 2024, 10:59am

How do I use the LLM Chain, can you post an example?

system · June 27, 2024, 11:00am

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.