#Clarification on Credit pricing in Open router

5 messages · Page 1 of 1 (latest)

cinder ember
#

Dear support,

I am looking to use Open router as the llm service provider for my current gen ai application. As part of it, I am looking for the prices and could not find the clarity on credit system in Open Router.
Could you please answer the below questions?

  1. How many credits will I receive per dollar?
  2. If I am using the llama 3.3 70b Instruct model, how will I be charged for 1request per second, and other cases like 5 requests per second, 10 requests per second etc.
  3. Is there any daily limit/ hourly limit on the number of requests for LLMs? especially llama 3.3 70b Instruct.
  4. If I am using llama 3.3 70b instruct, how much will I be charged in total at Open Router for 1million input tokens, 1 million output tokens, average of 10,000 requests per day, 10 requests parallely at same time?

Your response is highly appreciated and helps me choose the provider with ease.
Looking forward to hearing from you soon

Thanks and Regards
Dhanvanth Voona

low loom
# cinder ember Dear support, I am looking to use Open router as the llm service provider for m...

  1. "Credits" are another word for "dollar". So for every $1, you will receive 1 credit (this not accounting for fees. There's $0.40 flat fee for every deposit + 4% fee)

  2. Rate limits are based off how much money in your account.

    0.5 credits → 1 req/s (minimum)
    5 credits → 5 req/s
    10 credits → 10 req/s
    500 credits → 500 req/s
    1000 credits → Contact us if you see ratelimiting from OpenRouter

  1. No limit for paid LLMs, but there's a 200 request a day limit for :free LLMs

  2. Let me do the math. Gimme a sec

#
  1. Assuming the cheapest provider is used for all request (OpenRouter doesn't always use the cheapest provider, but you can force this behavior to hapapen if you want to. Let me know if you want to know how)

You will pay $0.12 cents for 1 million input tokens and $0.30 cents for 1 million output tokens. In total, you will pay $0.42 cents.

You are not charged (directly) for rate limits or parallel requests

cinder ember
#

Okay here is my understanding, so if I recharge 5 dollars initially, then I will be receiving 5 credits, that means I can use the model to start generating the text, while I have 5 credits, I can call 5 requests per second, if I exhaust 1 credit[ by using 3million tokens in Input and Output] then my balance would be reduced to a 3.76 credits, then I can only call 3 requests per second. And I wont be charged anything for calling mutliple requests per second.
Is my understanding correct? else correct me if I am wrong.

Also, help me choose How I can force to pick cheapest provider, if I find the results not accurate, then how can I change to service provider by Groq.

low loom
# cinder ember Okay here is my understanding, so if I recharge 5 dollars initially, then I will...
  1. You will not receive 5 credits because OpenRouter has deposit fees + Stripe (the payment processor) also has fees. Please let me know if you want more clearification
  2. Yes. If you have 3 credits. you can only send 3 requests per second.
  3. Groq is not listed on OpenRouter right now

To force OpenRouter to always choose the cheapest provider, add this to your request body (make sure to replace "OpenAI" with the actual provide you want to use)

{
  "provider": {
    "order": ["OpenAI"],
  },
  "allow_fallbacks": false
}

If you switch LLMs frequently, you can automate this by sending a API request to https://openrouter.ai/api/v1/models/[SLUG]/endpoints where [SLUG] is the ID of your model on OpenRouter and then going through the reply and determining what the cheapest provider is and replacing "OpenAI" with the name of the cheapest provider