Cost Engine
Prompt Cost Calculator
Stop staring at pricing tables. Paste your actual prompt, estimate your output, and see exactly how much your AI workload will cost across the industry's top foundation models at any scale.
1. Input Context (Prompt)
OR OVERRIDE
2. Output Size
tokens
3. Scale
times
Total Payload
500 tokens/run
Potential Savings
$7.36/mo
Model
Per Run
Monthly
DeepSeek V3DeepSeek
$0.0001
$0.14
GPT-4o-miniOpenAI
$0.0003
$0.30
Llama 3.1 70BMeta (Groq)
$0.0004
$0.40
Claude 3.5 HaikuAnthropic
$0.0006
$0.63
Gemini 1.5 ProGoogle
$0.0025
$2.50
GPT-4oOpenAI
$0.0075
$7.50
Claude 3.5 SonnetAnthropic
$0.0075
$7.50
How this engine works
Language models do not process text in words or characters; they process Tokens. Understanding the cost of a prompt requires estimating both the input (the context you provide) and the output (the generated response).
- The Token Heuristic: A standard rule of thumb for English text is that 1 token is approximately 4 characters, or about 0.75 words. This calculator uses this highly accurate heuristic to estimate the token count of pasted text in real-time.
- Input vs Output Economics: Models price generated (output) tokens significantly higher than processed (input) tokens. A heavy-RAG workload with massive input context but a short summary output is cheap. A creative writing workload with a short prompt and massive output is expensive.
- Scale: A prompt costing \.02 seems trivial, but if your application executes that prompt 100,000 times a month, the choice of model dictates whether your cloud bill is \ or \,000.