https://vapi.ai/ logo
Join Discord
Powered by
Call cost
# support
s
Hello, my call id 60289f96-f570-4075-a1c9-3151b40b378a and i want to know why 6 minustes of call is so expensive? 3.64$ for llm , how i can achieve lower price?
v
@San Sanych Hello! I understand you're concerned about the cost of a 6-minute call being $3.64. Let's break it down and see how we can potentially lower the cost for you. Based on the pricing information provided, here are the potential costs for a 6-minute call: - If the call was made using VAPI, the cost would be 6 minutes x $0.05/min = $0.30. - If the call was made using Twilio, the cost for an inbound toll-free call would be 6 minutes x $0.02/min = $0.12. - If the call was made using PlayHT, the cost would be 6 minutes x $0.07/min = $0.42. The cost you mentioned, $3.64, seems higher than the expected costs based on the provided rates. To help you achieve a lower price, we can look into the following: 1. Review the details of your call to ensure there are no additional services or features being charged. 2. Consider if there are any subscription plans or volume discounts that you might be eligible for. 3. Compare the costs of different providers to find the most cost-effective option for your needs. If you provide me with more details about the services you're using and any additional features that might be included in your calls, I can assist you further in understanding the costs and finding ways to reduce them. If the call ID you provided is related to a specific service, please let me know which one, so I can give you more accurate information.
You can continue this conversation by mentioning me in the message.
s
@User @User
Also i see that today many of my calls really costs so expensive. Earlier my calls costs me like 0.05 - 0.5$ per call, but now 0.84$ for 1:45 call? Yeasterday 2 minutes call costs 0.19$. Is it bug or what? https://cdn.discordapp.com/attachments/1305965315938848901/1305974082336063589/image.png?ex=6734faa5&is=6733a925&hm=e1ae1b81decbf774bf71bea6de560447f9315642232ace2c6b0d7b15432db552&
llm $3.65 ($3.65 / min) ‣ 767533 prompt tokens ‣ 1104 completion tokens it looks like bug
Hello??? can anybody help me with?
s
Hey @San Sanych here's the breakdown of the cost.
Copy code
md
- Total Cost: $4.2734

1. Transcriber: $0.0648
   - Model: `nova-2-general` (Provider: `deepgram`)
   - Usage: 6.55 minutes

2. Model: $3.6472
   - Model: `gpt-4o` (Provider: `openai`)
   - Usage: 767,533 prompt tokens and 1,104 completion tokens

3. Voice: $0.1955
   - Voice ID: `sRk0zCqhS2Cmv0bzx5wA` (Provider: `11labs`)
   - Characters: 3,909

4. VAPI: $0.3192
   - Type: `normal`
   - Usage: 6.38 minutes

5. Analysis (Summary): $0.0084
   - Model: `claude-3-5-sonnet-20240620` (Provider: `anthropic`)
   - Token Usage: 2,029 prompt tokens and 154 completion tokens

6. Analysis (Structured Data): $0.0065
   - Model: `claude-3-5-sonnet-20240620` (Provider: `anthropic`)
   - Token Usage: 2,064 prompt tokens and 18 completion tokens

7. Analysis (Success Evaluation): $0.0319
   - Model: `claude-3-5-sonnet-20240620` (Provider: `anthropic`)
   - Token Usage: 10,611 prompt tokens and 4 completion tokens
Hey @San Sanych the cost depends upon tokens, characters, model, voice used etc. There is no fixed cost concept even failed API calls are added as cost. As it's visible you have used 700k prompt tokens ad 1k completion tokens which summed up the total LLM cost. Let me know if you still have doubts.
s
But 1 day ago the same prompt have lowest price. @Shubham Bajaj
and at the end my account is frozen with -3$ balance) its crazy) last time i pay 20$ for 1-2 days, but this time it took like 20$+ with 30 calls like 1 min every call @Shubham Bajaj Okey, one thing that i changed it i is provider. i swapped from gpt 3.5 turbo to gpt 4o and as i see price increased like in 100 times, is it real?
@Shubham Bajaj as i understand, the one model that works good with functions IS gpt 4o? i tried all modes, but they all have problems, gpt4o one who works fine. i see that gpt 4mini model costs very cheap, but i see that functions still working bad. What you can advice me ? Is there some providers/models that works fine with functions along with system prompt and cost not like gpt4o?)
s
I advise to use GPT-4o it works the best compared to other models. GPT-40-mini for simple, redudant tasks can be used. Claude-3.5 may be a good choice to explore.
Yes Gpt-4o is costlier compared to gpt-3.5 which you can even see from the dashboard.
Hey @San Sanych If I can close your ticket now?
s
I think yes, closed Today all day testing gpt 4o and I see very good prices, not high as I saw previously, I think you just overdraft me with no reason, I guess. I have new tickets with new questions, waiting for response, thanks
18 Views
PreviousNext
Powered by