| Model | Availability | Input | Output | Monthly total | Price source |
|---|
| Model | Per second | Monthly total | Price source |
|---|
Monthly cost = input tokens not served from cache × input price + cached input tokens × cache-hit price + output tokens × output price. Token amounts are in millions because vendors price per 1M tokens.
DeepSeek publishes peak prices and off-peak prices at half. The calculator blends them using the peak share you enter.
DeepSeek, GLM and Kimi publish a cache-hit input price, so the cache-hit share is applied to them. No cache price is listed for Qwen in our checked data, so Qwen uses the full input price.
Qwen3.7-plus has a higher price when a request's input is between 256K and 1M tokens. Pick the tier that matches your typical request.
Wan is priced per second of generated video. Seedance 2.0 is billed by tokens; the per-second figure here is derived from ByteDance's own estimate example (5-second 16:9 clip, no video input).
| Model | Input | Cache-hit input | Output | Notes |
|---|---|---|---|---|
| deepseek-v4-pro | 1.32 | 0.044 | 3.96 | peak · api-docs.deepseek.com |
| deepseek-v4-pro | 0.66 | 0.022 | 1.98 | off-peak · api-docs.deepseek.com |
| deepseek-flash | 0.30 | 0.006 | 1.20 | peak · api-docs.deepseek.com |
| deepseek-flash | 0.15 | 0.003 | 0.60 | off-peak · api-docs.deepseek.com |
| qwen3.8-max | 2.00 | – | 6.00 | alibabacloud.com |
| qwen3.7-max | 2.50 | – | 7.50 | alibabacloud.com |
| qwen3.7-plus | 0.40 | – | 1.60 | input up to 256K · alibabacloud.com |
| qwen3.7-plus | 1.20 | – | 4.80 | input up to 1M · alibabacloud.com |
| glm-5.3 | 1.40 | 0.26 | 4.40 | docs.z.ai |
| kimi-k3.1-pro | 2.00 | 0.20 | 10.00 | platform.kimi.ai |
| kimi-k3.1-pro-highspeed | 4.00 | 0.20 | 20.00 | platform.kimi.ai |
| kimi-k3 | 3.00 | 0.30 | 15.00 | platform.kimi.ai |
| Seedance 2.0 | 480p 0.07 · 720p 0.152 · 1080p 0.374 · 4K 0.778 | ai.byteplus.com |
| wan3.0-video-prime | 480p 0.068 · 720p 0.14 · 1080p 0.28 | alibabacloud.com |
| wan3.0-video | 480p 0.05 · 720p 0.1 · 1080p 0.2 | alibabacloud.com |
| wan2.7-t2v / wan2.7-i2v | 720p 0.1 · 1080p 0.15 | alibabacloud.com |
| wan2.6-t2v / wan2.6-i2v | 720p 0.1 · 1080p 0.15 | alibabacloud.com |
USD per 1M tokens. Video: USD per second of generated video.
No. Every number is the vendor's official public list price, taken from the vendor's own pricing page and checked 2026-11-01. Our own price is quoted one-to-one based on your monthly volume; we do not publish a price list or discount levels.
How does the calculator handle DeepSeek peak and off-peak hours?DeepSeek lists a peak price (01:00–04:00 and 06:00–10:00 UTC, Monday to Friday) and an off-peak price at half. Enter the share of your traffic that falls in peak hours and the calculator blends the two.
What does the cache-hit percentage change?It is the share of your input tokens billed at the vendor's cache-hit input price. DeepSeek, GLM and Kimi publish one; for Qwen the calculator uses the full input price.
Which of these models can I call through your OpenAI-compatible endpoint?deepseek-v4-pro, deepseek-flash, qwen3.8-max, qwen3.7-plus and glm-5.3. Kimi, Seedance 2.5 and Wan are supplied by quote. API keys are issued after your quote is paid.
How do I get a price for my volume?Click “Get a quote for this volume”: the quote form opens with your volume and models filled in. You can also ask Ray, our AI assistant, or message Sapporo on Telegram @sapporo168. We reply within 24 hours.