Cost calculator
Describe your workload and see what every model would cost per month, cheapest first. Prices are USD from vendor price lists; quality filters use heuristic task scores, so shortlist here and then test on your own data.
Start from an example
≈ 50,000 requests per month
per request
per request
Repeated prompt parts (system prompt, documents). Only models with a cached-input price benefit.
Work that can wait hours. Only models with batch prices benefit; batch and cache discounts are not stacked.
Showing only the 1 model you picked (quality filters are ignored).
1 of 490 models · hidden: 489 by filters
Cheapest
$22.50/ month
Qwen3 235B A22B Instruct 2507
$0.00045 per request
Median
$22.50/ month
Qwen3 235B A22B Instruct 2507
$0.00045 per request
Most expensive
$22.50/ month
Qwen3 235B A22B Instruct 2507
$0.00045 per request
Spread
1×
priciest vs cheapest
$0 / month apart
| Compare | # | Model | Per request | Per month | vs cheapest | Discounts used | Context | Price in / out |
|---|---|---|---|---|---|---|---|---|
| 1 | Qwen3 235B A22B Instruct 2507 Alibaba (Qwen)fast | $0.00045 | $22.50 | cheapest | — | 131K | $0.14 / $0.80 |
Showing 1–1 of 1
Cost = input tokens × input price + output tokens × output price, per request, times requests per month. Excludes taxes, volume discounts, fine-tuning, tool-call fees and any minimum commitments. Tick “Compare” on up to four rows to see them side by side.