Input tokens overhead

[usatenko]

from openai import OpenAI
client = OpenAI()

response = client.responses.create(
    model="computer-use-preview",
    tools=[{
        "type": "computer_use_preview",
        "display_width": 1024,
        "display_height": 768,
        "environment": "browser" # other possible values: "mac", "windows", "ubuntu"
    }],
    input=[
        {
            "role": "user",
            "content": "Check the latest OpenAI news on chrome.com."
        }
        # Optional: include a screenshot of the initial state of the environment
        # {
        #     type: "input_image",
        #     image_url: f"data:image/png;base64,{screenshot_base64}"
        # }
    ],
    reasoning={
        "generate_summary": "concise",
    },
    truncation="auto"
)

print(response)

for this code I get the answer:

... status='completed', text=ResponseTextConfig(format=ResponseFormatText(type='text')), truncation='auto', usage=ResponseUsage(input_tokens=990, output_tokens=7, output_tokens_details=OutputTokensDetails(reasoning_tokens=0), total_tokens=997, input_tokens_details={'cached_tokens': 0}), user=None, store=True)

So, if I even do not send a picture, it anyway uses 990 input tokens. With a picture it is up to 3000-4000 tokens.
To process 30 screenshot iterations by this model by the agent it takes 30 * 1000 = 30000 of input tokens overhead (if $3/1 mln tokens = $0.09).
Is it correct?

[katia-openai]

Thanks for reporting this, this is an issue we're aware of and we're currently looking into. Will keep you posted!

[Everything-In-Sync]

@katia-openai Hey Katia, I ran it again today and this time it did send over the base64 from the screenshot (im not going to copy and paste all of that text here) which it did not do last time so that part is working, it's just the same output issue from the model: {'error': {'message': 'The model computer-use-preview-2025-03-11 does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

Thank you for the progress

[theonepa1]

@Everything-In-Sync - ran into same issue. Looks like the model no longer exists.