Date: Nov 12th 2024
- A random prompt and output was saved and recorded (
/examples/1.md) - The access costs of major cloud-hosted LLM APIs were gathered and recorded using data from the providers themselves.
- The data was gathered on 12-Nov-24 and input and output tokens were captured as they were expressed (US dollars per one million tokens USD/MTok). For the purpose of simplicity, discounts were ignored (some providers offer discounts for routing prompts to a random endpoint, for example)
- The sample prompt and output were tokenised using a tokenisation calculator. For Cohere, an average of the other tokenisation values was used.
The data was combined and calculated as follows:
- The calculated token usage for inputs and outputs was divided by one million
- That number was multipled by the MTok values for input and output tokens respectively.
- Then, the numbers were added to give a total cost for this sample prompt and output (
1.md). - Finally, 20 was divided by this number and the result was rounded to the nearest integer.
- This number represents how many of the sample prompts could be carried out at $20 of expenditure.
- Finally, that figure was then divided by 30 to estimate how many prompts-and-outputs per day such as prompting budget would afford in a typical calendar month.
| LLM | Total Cost ($) | Per $20 | Per-Day |
|---|---|---|---|
| GPT 4o | 0.009328 | 2144 | 71 |
| GPT 4o mini | 0.000618 | 32378 | 1080 |
| Mistral Large | 0.006380 | 3135 | 105 |
| Claude 3.5 | 0.014664 | 1364 | 46 |
| Claude 3 Opus | 0.087660 | 228 | 7.6 |
| Cohere R+ | 0.009328 | 2144 | 71 |
| Cohere R | 0.0011193 | 17868 | 596 |
Based on a random prompt and output pair and using the LLMs' API pricing data available on Nov 12th 2024, how many prompts and outputs could you run every day before you reached the $20 a monthly ChatGPT sub costs?
.... how many would that work out to per day of a 30 day calendar month?
| Provider | Model | Input Price (USD/Mtok) | Output Price (USD/Mtok) |
|---|---|---|---|
| OpenAI | Gpt-40 | 2.50 | 10 |
| OpenAI | gpt-4o-2024-08-06 | 2.50 | 10 |
| OpenAI | GPT-4o mini | 0.15 | 0.6 |
| OpenAI | chatgpt-4o-latest | 5.00 | 15 |
| Anthropic | Claude 3.5 Sonnet | 3.00 | 15 |
| Anthropic | Claude 3.5 Haiku | 1.00 | 5 |
| Anthropic | Claude 3 Opus | 15.00 | 75 |
| Mistral | Large 2 | 2.00 | 6 |
| Mistral | Small 24.09 | 0.20 | 0.6 |
| Mistral | Codestral | 0.20 | 0.6 |
| Gemini 1.5 Flash (Paid Tier) | 0.07 | 0.30 | |
| Gemini 1.5 Pro (Paid Tier) | 1.25 | 5 | |
| Cohere | Command R+ | 2.50 | 10 |
| Cohere | Command R | 0.30 | 1.2 |