Cost formula
Cost for 1M input tokens = input price per 1M; Cost for 1M output tokens = output price per 1M; Tokens affordable = budget ÷ price per 1M × 1,000,000
Calculator
1 million tokens cost calculator
Pick an input/output mix (and optional cache or batch), then rank models by the cost of exactly one million tokens—and see how many tokens a fixed budget buys.
1M token reference
Gemini 2.0 Flash-Lite
$0.102for 1M tokens at this mix
$0.000102 / 1K · Google
Cheapest 1M$0.102
Priciest in set$240.00
Models ranked71
Tokens @ $100.00980,392,156
Mix80% in / 20% out
Spread (hi − lo)$239.90
Reference
1M token mix
Pick an input/output/cache mix, then compare list prices across models.
Remaining 20% billed as output
Compare
Model cost strip
Cheapest at this mix: Gemini 2.0 Flash-Lite
1$0.102
Gemini 2.0 Flash-Lite
2$0.1056
GPT-5 nano
3$0.1056
GPT-5.4 nano
4$0.1241
DeepSeek V4 Flash
5$0.1312
Gemini 3.1 Flash-Lite
6$0.136
Gemini 2.5 Flash-Lite
7$0.136
GPT-4.1 nano
8$0.136
Gemini 2.0 Flash
Showing top 8 of 71 ranked models. Spread from cheapest to priciest: $239.90.
“1M tokens” here means one million tokens under your chosen input/output/cache mix—not 1M input plus 1M output unless the mix is set that way. Updated 2026-07-31. Approximate static list prices for planning only. Always verify against each provider’s official pricing page before production budgeting.
Guide
How to get value from this calculator
See exactly how much one million input or output tokens costs by model, and invert the math to find how many tokens your budget buys. Useful for quick mental math when reading provider pricing pages or comparing per-million list rates across vendors.
Why it matters
Provider pricing is published per million tokens, but product teams think in requests and budgets. This calculator bridges list prices and actionable volume planning without manual arithmetic.
How to use it
- Select a model or browse the comparison list.
- Review input cost per 1M tokens and output cost per 1M tokens separately.
- Enter a dollar budget to see how many input or output tokens it purchases.
- Compare the same 1M-token cost across two or three candidate models.
- Remember output tokens often cost more—price your workload mix, not input alone.
- Verify rates on the provider's official pricing page before committing.
Planning tips
- For mixed workloads, weight input and output 1M costs by your actual token ratio.
- Batch pricing applies a discount to the per-1M rate—check batch columns if eligible.
- Cached input reads use a lower effective per-1M rate—adjust for cache hit rate when relevant.
- A $100 budget buys vastly different output tokens depending on model—compare before defaulting to flagship.
- Scan the ranked model strip first, then open the price table for deeper filters.
- Use alongside the price table for multi-model scans.
Frequently asked questions
Does 1M tokens mean input plus output combined?
Input and output are priced separately. One million input tokens and one million output tokens have different costs on most models.
Are these per-million rates official?
CentsPerToken uses approximate list prices for planning. Providers update rates—confirm officially before budgeting.
How many words is 1M tokens?
Roughly 750,000 words in English for many models, but varies by language and content. Use the token counter for precise counts.
Can I use this for embedding tokens?
Embedding models have their own per-1M rates. Select an embedding model if available or use the embedding RAG calculator for pipeline context.