#Prices OpenAI #2

1 messages · Page 1 of 1 (latest)

stone idol
#

Lets say I have a budget of 50 dollar, I send requests to the chat API on friday, saturday and sunday on a weekly basis. How many requests per hour can I do?

zenith bridgeBOT
#

Hi @stone idol. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

To calculate the number of requests you can make to the OpenAI API within your budget, you need to consider the cost per token for the model you're using. According to the information provided in the extracts, the cost per 1K tokens for the gpt-3.5-turbo model is $0.0010 for input and $0.0020 for output.

However, the number of tokens used per request can vary greatly depending on the length and complexity of the input and output. Therefore, it's not possible to provide an exact number of requests you can make per hour with a $50 budget without knowing the average number of tokens used per request.

To stay within your budget, you could monitor your usage and adjust the frequency of your requests accordingly. If you're using AWS services, as suggested in one of the forum posts, you could use Amazon SQS and AWS Lambda to manage your requests and control concurrency. This could potentially allow you to stay within the free tier usage of the Lambda functions, depending on the resources allocated to each function and the time taken for each call to run.

For more detailed information on pricing and usage, you can visit the OpenAI Pricing Page (https://openai.com/pricing) and the OpenAI API documentation (https://platform.openai.com/docs).

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@stone idol kapa.ai is still learning and improving, please let us know how it did by reacting below

stone idol
#

@zenith bridge Ah my appologies lets say average of 250 tokens. Also operational hours are 7 AM to 11 PM

zenith bridgeBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Given your budget of $50, the average tokens per request being 250, and your operational hours from 7 AM to 11 PM, we can calculate the number of requests you can make.

First, let's calculate the cost per request. If the cost per 1K tokens is $0.0030 (sum of $0.0010 for input and $0.0020 for output), then the cost for 250 tokens would be $0.0030 * 250 / 1000 = $0.00075 per request.

Next, let's calculate the total number of requests you can make within your budget. If your budget is $50, then the total number of requests you can make is $50 / $0.00075 = 66,667 requests.

Now, let's calculate the number of operational hours per week. If your operational hours are from 7 AM to 11 PM, that's 16 hours per day. Over three days (Friday, Saturday, Sunday), that's 16 * 3 = 48 hours.

Finally, let's calculate the number of requests per hour. If you can make 66,667 requests over 48 hours, then the number of requests per hour is 66,667 / 48 = approximately 1,389 requests per hour.

Please note that this is a rough estimate and actual usage may vary. Also, remember to consider the rate limits imposed by OpenAI. For example, the rate limit for the gpt-3.5-turbo model is 60 RPM (requests per minute) or 3,600 RPH (requests per hour) for free trial users, and 3500 RPM or 210,000 RPH for pay-as-you-go users.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@stone idol kapa.ai is still learning and improving, please let us know how it did by reacting below

stone idol
#

@zenith bridge I got 50 dollars in a month, so lets say its on a monthly basis

zenith bridgeBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Given your budget of $50 per month and an average of 250 tokens per request, we can calculate the number of requests you can make in a month.

First, let's calculate the cost per request. If the cost per 1K tokens is $0.0030 (sum of $0.0010 for input and $0.0020 for output), then the cost for 250 tokens would be $0.0030 * 250 / 1000 = $0.00075 per request.

Next, let's calculate the total number of requests you can make within your budget. If your budget is $50, then the total number of requests you can make is $50 / $0.00075 = 66,667 requests.

Now, let's calculate the number of operational hours per month. If your operational hours are from 7 AM to 11 PM, that's 16 hours per day. Over a month (30 days), that's 16 * 30 = 480 hours.

Finally, let's calculate the number of requests per hour. If you can make 66,667 requests over 480 hours, then the number of requests per hour is 66,667 / 480 = approximately 139 requests per hour.

Please note that this is a rough estimate and actual usage may vary. Also, remember to consider the rate limits imposed by OpenAI. For example, the rate limit for the gpt-3.5-turbo model is 60 RPM (requests per minute) or 3,600 RPH (requests per hour) for free trial users, and 3500 RPM or 210,000 RPH for pay-as-you-go users.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@stone idol kapa.ai is still learning and improving, please let us know how it did by reacting below

stone idol
#

Alright, lets say I do 60 requests per hour during those operation hours and days, how many tokens can I use on average?

#

@zenith bridge Alright, lets say I do 60 requests per hour during those operation hours and days, how many tokens can I use on average?

zenith bridgeBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Given your budget of $50 per month, your operational hours from 7 AM to 11 PM on Friday, Saturday, and Sunday, and a rate of 60 requests per hour, we can calculate the average number of tokens you can use per request.

First, let's calculate the total number of requests you can make within your budget. If your budget is $50 and the cost per 1K tokens is $0.0030 (sum of $0.0010 for input and $0.0020 for output), then the total number of tokens you can use is $50 / $0.0030 * 1000 = 16,666,667 tokens.

Next, let's calculate the total number of operational hours per month. If your operational hours are from 7 AM to 11 PM, that's 16 hours per day. Over a month (4 weeks), that's 16 * 3 * 4 = 192 hours.

Then, let's calculate the total number of requests per month. If you're making 60 requests per hour for 192 hours, then the total number of requests is 60 * 192 = 11,520 requests.

Finally, let's calculate the average number of tokens per request. If you can use 16,666,667 tokens over 11,520 requests, then the average number of tokens per request is 16,666,667 / 11,520 = approximately 1,446 tokens.

Please note that this is a rough estimate and actual usage may vary. Also, remember to consider the rate limits imposed by OpenAI. For example, the rate limit for the gpt-3.5-turbo model is 60 RPM (requests per minute) or 3,600 RPH (requests per hour) for free trial users, and 3500 RPM or 210,000 RPH for pay-as-you-go users.

#

@stone idol kapa.ai is still learning and improving, please let us know how it did by reacting below

stone idol
#

@zenith bridge what if I want to use GPT 4?

zenith bridgeBOT
#

Let me check the knowledge base for you. One moment! :books:

#

If you want to use GPT-4, the rate limits and token usage would depend on the specific details of the GPT-4 model, which are not provided in the extracts. However, some general information about rate limits and usage tiers for GPT-4 can be found in the extracts.

For instance, the rate limits for the GPT-4 preview model are fixed at 20 requests per minute and 100 requests per day, and it's not possible to increase these limits for the preview model. A stable model will become available in the coming weeks.

In terms of usage tiers, there are tokens per minute (TPM) limitations based on how much you have already paid to use the API. For example, if you have paid $50 and it's been 7+ days since your first successful payment, you would be in Tier 2, which has a limit of 40,000 TPM.

However, to calculate the number of requests you can make or the average number of tokens you can use per request with GPT-4, we would need more specific information about the cost per token and the average tokens per request for GPT-4, which are not provided in the extracts.

For more detailed information on GPT-4 rate limits and usage tiers, you can check the Limits section (https://platform.openai.com/account/limits) of your account settings and the usage tiers guide (https://platform.openai.com/docs/guides/rate-limits/usage-tiers).

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@stone idol kapa.ai is still learning and improving, please let us know how it did by reacting below

stone idol
#

@unreal vector it knows the cost of the GPT 4 models correct? So did it just fail to find the prices here?

runic shell
#

@zenith bridge Can someone use chagpt ( api ) instead of chatgpt plus with the same budget or it will be more expensive?

zenith bridgeBOT
#

Let me check the knowledge base for you. One moment! :books:

#

@runic shell That's a great question, but unfortunately I am only capable of responding to a maximum of five questions per thread. Don't hesitate to create a new thread for any further questions you have though. I'm here to help!

unreal vector
#

@stone idol just confirmed that it does know prices here: #1184563609842487488 message 👋

I think this is a case of hitting the limits of the models when we're 4 questions deep in a reasoning chain.