Please help me to understand text search api usage and charges.
Also please provide comparation between chat gpt and gemini ?
As gemini price is less then how you can explore chat gpt is better
Hi @govindzeffy and welcome to the community!
Gemini is accessed via Google AI Studio and you can see the pricing for Gemini 1.5 Pro here. If you are a large enterprise then you usually get it via Google’s Vertex AI platform, and not the AI Studio.
OpenAI pricing is clearly stipulated here.
It’s important to note that for API, it’s pay-as-you-go model. For ChatGPT it’s a per-seat pricing (you pay a monthly subscription).
If I compare them side-by-side (context window of 128K tokens):
- Input tokens:
- Gemini 1.5 Pro: $1.25 / 1 million tokens
- GPT-4o: $2.50 / 1 million tokens
- Output tokens:
- Gemini 1.5 Pro: $5.00 / 1 million tokens
- GPT-4o: $10.00 / 1 million tokens
But there are some very important points worth mentioning here:
- GPT-4o is a more performant model than Gemini (see here for the latest leaderboard)
- The OpenAI ecosystem is much richer in the sense that you can create Assistants, custom GPTs, and have some really cool API features like Realtime API
- The pricing can be significantly lower due to the use of prompt caching and Batch API
Hope that helps!
Thank you so much @platypus . Its really helpful.
Could you please help me to understatnd about
what is input token ?
what is ouput token ?
can you please explain by my example
If i searching
"Eladeca is valid company "
how much input token i am spending here ? and how output token will work here ?
Happy to help!
Input tokens are what goes into your prompt, which is usually some kind of a system prompt, e.g. “You are a helpful assistant”, and your instruction or a question, and optionally any background/knowledge context.
Output tokens are what GPT-4o returns.
In English there are on average around 1,3 tokens per word, but exact number will vary. You can play around with this tokenization here.
So if we use your example, let’s say your system prompt is “You are a helpful assistant”, and your user prompt is simply “Eladeca is valid company”, this would equate to 12 input tokens, which would set you back 0.003 cents!
In my response/output, I got the following:
Eladeca Private Limited is a legitimate IT services provider based in Noida, Uttar Pradesh, India. The company was incorporated on May 27, 2023, and is registered with the Ministry of Corporate Affairs in India. Eladeca specializes in various IT solutions, including web and mobile application development, catering to industries such as e-commerce, online gaming, healthcare, and education. Their official website provides detailed information about their services and portfolio.
So this is 90 tokens, which would cost you 0.09 cents.