Formula Input Tokens × Input $/1M ÷ 1,000,000 + Output Tokens × Output $/1M ÷ 1,000,000 = Cost per Query
Step 1 — Select or Configure Model
Custom / Other Model manual entry
$ / 1M tok
$ / 1M tok
Step 2 — Input Tokens (Prompt)
tokens
010k25k50k
📌 ~1 token ≈ 4 chars in English. A typical user prompt is 50–500 tokens. System prompts add 200–2000 tokens on top.
Paste text to estimate tokens
Estimated tokens: 0  |  Characters: 0
Step 3 — Output Tokens (Response)
tokens
04k8k16k
📌 A short reply is ~100–300 tokens. A detailed essay is ~1000–3000 tokens. Code with explanation: ~500–2000 tokens.
Paste expected response to estimate
Estimated tokens: 0  |  Characters: 0
Cost Breakdown
INR per $1
Update this to today’s FX rate.
Input Cost
$0.000000
— tokens × —
Output Cost
$0.000000
— tokens × —
Total Per Query
$0.000000
— queries per $1
Total Per Query (INR)
INR 0.0000
— queries per INR 1
Model—
Input tokens—
Input price ($/1M)—
Input cost—
Input cost (INR)—
Output tokens—
Output price ($/1M)—
Output cost—
Output cost (INR)—
Total per query—
Total per query (INR)—
Scale Projections
Configure Query
tokens
tokens
All Models — Cost Comparison
Model Provider Input $/1M Output $/1M Cost/Query Queries/$1
What is a Token?

Tokens are chunks of text — roughly 4 characters in English, or about ¾ of a word. Tokenization splits words into subword units.

Examples:

"Hello world"2 tokens
"OpenAI"2 tokens
"Anthropomorphic"4 tokens
1 sentence (~60 chars)~15 tokens
1 page of text (~500 words)~650 tokens
1 novel (~90k words)~120k tokens

Code tends to use more tokens than prose due to symbols and whitespace.

Quick Estimation Rules
Rule of thumb1 token ≈ 4 chars
Chars → tokenschars ÷ 4
Words → tokenswords × 1.33
Tokens → wordstokens × 0.75
Short chat message10–50 tokens
Typical prompt100–500 tokens
System prompt200–2000 tokens
Short AI reply100–300 tokens
Long detailed reply1000–3000 tokens
Code file (~200 lines)~500–1000 tokens
Pricing Models Explained
Per-Token Pricing (most common)
Charged separately for input and output tokens.
Formula: (input_tok ÷ 1M × input_price) + (output_tok ÷ 1M × output_price)
Batch Pricing
Many providers offer 50% discount for async/batch requests that don't need real-time responses. Great for bulk processing.
Context Caching
Anthropic & Google offer caching of repeated context (system prompts, docs). Cached reads are ~90% cheaper. Write cache once, save on reads.
Subscription Plans
Some providers charge flat monthly fees. Better if usage is predictable and high volume. Calculate break-even vs pay-per-token.