Converter
Tokens to dollars converter
Pick a model, enter an input and output token count, see the exact dollar cost. Uses the live Calcis pricing table — no stale rate cards.
Pick a model, enter token counts, see the dollar cost
Input cost
$0.00250
Output cost
$0.00750
Total
$0.01000
Input: $2.50/ 1M · Output: $15.00/ 1M
How the conversion works
LLM API cost is simply (input tokens × input rate) + (output tokens × output rate), where rates are quoted per 1 million tokens. A 1,000-token input on GPT-5 ($2.50/1M) costs $0.0025; a 500-token output ($15/1M) costs $0.0075.
The converter below pulls its rates from the live Calcis pricing table and also applies tokenizer multipliers (Claude Opus 4.7 rebills at 1.15× the token count) and long-context tiers (Gemini Pro doubles its rate above 200K input tokens). That makes the number you see match what you'll actually be billed.
For cached inputs (Anthropic prompt caching, OpenAI automatic caching), apply the cached-input rate manually — this converter uses base rates to avoid over-promising savings you haven't configured.
Cost landmarks at $2.50 input / $15 output (GPT-5)
| 1,000 input tokens | $0.0025 |
| 1,000 output tokens | $0.0150 |
| 10,000 tokens (5K in / 5K out) | $0.0875 |
| 100,000 tokens (50K in / 50K out) | $0.8750 |
| 1,000,000 tokens (500K / 500K) | $8.75 |
| A 30-page paper summary (24K in / 500 out) | $0.068 |
| A typical chat turn (700 in / 300 out) | $0.006 |
Frequently asked
Why do the rates differ between input and output?
Output tokens cost 4-10× more because generation is inference-heavy — each output token requires a full forward pass through the model, whereas input tokens are processed in parallel. This is true across every provider.
Does the calculator include cached-input discounts?
No — it uses the base rate so the number represents a worst-case read of the rate card. In practice, Anthropic's prompt cache cuts input cost 90%, OpenAI's cuts 50-75%, and Gemini's context caching cuts 75%. Apply the discount manually for your cache-hit rate.
What's the tokenizer multiplier doing?
Claude Opus 4.7 shipped a retrained tokenizer that emits ~15% more tokens than Opus 4.6 for the same text. Calcis applies a 1.15× multiplier to Opus 4.7 estimates so the cost you see matches your actual bill. No other tracked model has an active multiplier.
How does the Gemini long-context tier work?
Gemini 2.5 Pro charges base rates ($1.25 input, $10 output per 1M) up to 200K input tokens, and double rates ($2.50 / $15) above that. If your input is over 200K the converter applies the upper tier automatically.
What about batch API discounts?
OpenAI and Anthropic both offer 50% off on batch API requests (24-hour async delivery). This converter shows real-time pricing. For batch workloads, divide the shown cost by 2 — simple and correct.
Ready to estimate a real prompt?
Paste your actual text into the estimator for exact token counts and dollar costs across every model.