AI API Cost Per Request
Calculate the exact cost of a single API request for any model. See which ones cost pennies — and which cost dollars.
Quick Presets
$0.00188
cost per request on GPT-5
At 1,000 requests/day = $1.88/day = $56.25/month
All Models Ranked by Cost Per Request
Based on your input: 500 input + 200 output tokens
| # | Model | Provider | Input/1M | Output/1M | Cost/Request | Cost/1K Req | Savings vs GPT-5 |
|---|
💡 Why Per-Request Cost Matters
Token pricing is abstract. Per-request cost tells you what you actually pay. A chatbot handling 10K daily requests at $0.01/req costs $3,000/month — while the same workload on DeepSeek V4 Flash costs $84/month.
📊 The 470x Gap
The cheapest model (GPT-oss 20B) costs 245x less per request than the most expensive (GPT-5.5 Pro). Choosing the right model for your workload is the single biggest cost lever.
Find Your Cheapest Model
APIpulse Pro shows every alternative ranked by cost for your exact workload — with migration code to switch in 5 minutes.
⚡ Calculate Your Savings →Free calculator · No signup · See results instantly
Frequently Asked Questions
How do I calculate AI API cost per request?
Multiply input tokens by the input price per token, output tokens by the output price per token, then add them. For example, 1,000 input + 500 output tokens on GPT-5 ($1.25/$10.00 per 1M): (1000 × $1.25/1M) + (500 × $10.00/1M) = $0.00625 per request.
What is the cheapest AI API per request?
GPT-oss 20B at $0.08/$0.35 per 1M tokens — a typical 500/200 token request costs $0.00011. DeepSeek V4 Flash ($0.14/$0.28) is $0.000126. Both are under 1/100th of a cent per request.
How much does a GPT-5 API request cost?
GPT-5 costs $1.25/$10.00 per 1M tokens. A typical chat request (1,000 input + 500 output) costs $0.00625. At 100K requests/month, that's $625/month. Switching to DeepSeek V4 Flash cuts that to $12.60/month.
How much does a Claude API request cost?
Claude Sonnet 4.6 ($3.00/$15.00 per 1M) costs $0.0105 per typical request. Claude Haiku 4.5 ($1.00/$5.00) costs $0.0035 — 67% cheaper. For simple tasks, Haiku is the better value.
How do I reduce my AI API cost per request?
5 strategies: (1) Use cheaper models for simple tasks — DeepSeek V4 Flash at $0.00013/req vs GPT-5 at $0.006/req. (2) Cache repeated prompts. (3) Shorten system prompts. (4) Use structured output for shorter responses. (5) Batch requests when latency allows.
Does context window size affect cost per request?
Yes. Longer context = more input tokens = higher cost. A RAG query with 10K context tokens costs 10x more than a 1K chat message. Use retrieval to keep context under 2K tokens when possible.