ChuhaiStack

Home首页 / China AI中国 AI / API pricesAPI 价格

Chinese AI model API pricing tracker中国大模型 API 价格追踪

What do DeepSeek, Qwen, Kimi, GLM, MiniMax and ByteDance Seed cost via API? USD per 1M input/output tokens, copied from each provider's official international pricing page. 36 models, checked 2026-10-15.DeepSeek、千问、Kimi、智谱 GLM、MiniMax、字节 Seed 的 API 价格，单位为每百万 token 美元，全部摘自官方国际版价格页。共 36 个模型，核查于 2026-10-15。

This data is not sponsored and this page contains no affiliate links. We only list USD prices stated directly on official pages, with no currency conversion. Anything not stated is marked "Unknown" or "Not stated". Prices change often; confirm on the official page before you buy.本页数据不接受赞助，也不含联盟链接。只收录官方页面直接给出的美元价格，未做汇率换算；官方未写明的一律标为「未知」或「未注明」。价格随时可能变化，下单前请以官方页面为准。

Most providers also have cache, batch, peak/off-peak or long-context tiers; see the notes under each table.提示：多数厂商另有缓存、批量（Batch）、高峰/非高峰或长上下文分档价格，详见各表备注。

Official APIs官方 API

DeepSeek API (official) · DeepSeek

Prices shown are peak-hour list prices. Off-peak rates are half. Per DeepSeek, peak hours are 01:00-04:00 and 06:00-10:00 UTC, Monday-Friday (excluding Chinese public holidays); all other hours are off-peak.表中为高峰时段价格，非高峰时段为半价。据官方说明，高峰为周一至周五 UTC 01:00-04:00 和 06:00-10:00（中国法定节假日除外），其余时间均为非高峰。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
deepseek-flash (DeepSeek-V4.1-Flash)$0.30$0.006$1.201MOff-peak: $0.15 in / $0.60 out.
deepseek-v4-pro (DeepSeek-V4-Pro-0813)$1.32$0.044$3.961MOff-peak: $0.66 in / $1.98 out.

Source来源: https://api-docs.deepseek.com/quick_start/pricing · Checked核查于 2026-10-15 · Availability outside China海外可用性

Alibaba Cloud Model Studio (international) · Alibaba (Qwen)

Prices shown are for the "International" deployment scope (Singapore region). Model Studio also lists "Global", "US" and "EU" deployment scopes with different prices; see the official page. Several models use tiered pricing: the tier is set by the input tokens in a single request, and all tokens in that request are billed at that tier.表中为「International」部署范围（新加坡地域）的价格。百炼国际版另有「Global」「US」「EU」部署范围，价格不同，以官方页面为准。部分模型按单次请求的输入 token 数分档计价，整次请求按所在档位计费。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
qwen3.8-max$2.00Unknown未知$6.00Not stated价格页未注明≤1M input tokens/request
qwen3.7-plus$0.40Unknown未知$1.60Not stated价格页未注明≤256K input tokens/request · List price; the page shows a limited-time 20% discount. 256K-1M tier: $1.20 in / $4.80 out.
qwen3.8-flash$0.15Unknown未知$0.47Not stated价格页未注明≤1M input tokens/request
qwen3.7-flash$0.03Unknown未知$0.13Not stated价格页未注明≤32K input tokens/request · 32K-256K: $0.10 / $0.40; 256K-1M: $0.20 / $0.80.
qwen3-coder-flash$0.30Unknown未知$1.50Not stated价格页未注明≤32K input tokens/request · Higher tiers up to 1M: $0.50/$2.50, $0.80/$4, $1.60/$9.60.
qwen3.8-27b$0.50Unknown未知$3.00Not stated价格页未注明≤1M input tokens/request

Source来源: https://www.alibabacloud.com/help/en/model-studio/model-pricing · Checked核查于 2026-10-15 · Availability outside China海外可用性

Kimi API Platform (official) · Moonshot AI (Kimi)

Pay-as-you-go, no subscription. Prices exclude applicable taxes. kimi-k3 also bills cache writes separately ($3.00 per 1M for the 5-minute TTL, $6.00 for 1-hour TTL).按量付费，无订阅；价格不含税。kimi-k3 另按缓存写入计费（5 分钟 TTL $3.00/百万，1 小时 TTL $6.00/百万）。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
kimi-k3$3.00$0.30$15.001,048,576 tokens
kimi-k2.7-code$0.95$0.19$4.00262,144 tokens
kimi-k2.7-code-highspeed$1.90$0.38$8.00262,144 tokens
kimi-k2.6$0.95$0.16$4.00262,144 tokens

Source来源: https://platform.kimi.ai/docs/pricing/chat · Checked核查于 2026-10-15 · Availability outside China海外可用性

Z.ai API (international) · Zhipu AI (GLM)

Z.ai states all prices are in USD. Cached-input storage is listed as limited-time free. Context length is not stated on the pricing page.Z.ai 注明所有价格为美元；缓存存储费标注为限时免费。价格页未写明上下文长度。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
GLM-5.3$1.40$0.26$4.40Not stated价格页未注明
GLM-5.3-FlashX$0.37$0.075$1.25Not stated价格页未注明
GLM-5.3-Flash$0.15$0.03$0.50Not stated价格页未注明
GLM-5.1$1.40$0.26$4.40Not stated价格页未注明
GLM-5$1.00$0.20$3.20Not stated价格页未注明
GLM-4.7$0.60$0.11$2.20Not stated价格页未注明
GLM-4.7-FlashFree免费Free免费Free免费Not stated价格页未注明Listed as free.

Source来源: https://docs.z.ai/guides/overview/pricing · Checked核查于 2026-10-15 · Availability outside China海外可用性

MiniMax API Platform (international) · MiniMax

MiniMax-M3 prices shown are after a discount the page labels "Permanent 50% off" (list $0.60 in / $2.40 out for ≤512K). The page also lists a Priority tier at 1.5x standard pricing. Context length is not stated on the pricing page.MiniMax-M3 为官方标注「永久 5 折」后的价格（≤512K 原价 $0.60 输入 / $2.40 输出）；另有 Priority 档，价格为标准的 1.5 倍。价格页未写明上下文长度。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
MiniMax-M3$0.30$0.06$1.20Not stated价格页未注明≤512K input tokens · >512K input: $0.60 in / $2.40 out.
MiniMax-M2.7$0.30$0.06$1.20Not stated价格页未注明
MiniMax-M2.7-highspeed$0.60$0.06$2.40Not stated价格页未注明
MiniMax-M2.5$0.30$0.03$1.20Not stated价格页未注明

Source来源: https://platform.minimax.io/docs/guides/pricing-paygo · Checked核查于 2026-10-15 · Availability outside China海外可用性

BytePlus ModelArk (international) · ByteDance (Seed)

Online inference (standard) prices. Seed models are tiered by prompt length (K tokens). The page also lists Flex and Batch prices, roughly half for some models. Context length is not stated on the pricing page.表中为在线推理（标准）价格。Seed 模型按提示词长度（K tokens）分档；页面另列 Flex 和 Batch 价格（部分模型约为一半）。价格页未写明上下文长度。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
seed-2-0-pro-260328$0.50$0.10$3.00Not stated价格页未注明Prompt ≤128K · 128K-256K: $1.00 in / $6.00 out.
seed-2-0-lite-260428$0.25$0.05$2.00Not stated价格页未注明Prompt ≤128K · 128K-256K: $0.50 / $4.00. Audio input priced separately.
seed-2-0-mini-260428$0.10$0.02$0.40Not stated价格页未注明Prompt ≤128K · 128K-256K: $0.20 / $0.80. Audio input priced separately.
dola-seed-2-1-turbo$0.50$0.10$2.50Not stated价格页未注明Prompt ≤256K
seed-2-0-code-preview-260328$0.50$0.10$3.00Not stated价格页未注明Prompt ≤128K · 128K-256K: $1.00 / $6.00.

Source来源: https://ai.byteplus.com/ark/region:ap-southeast-1/docs/model-pricing · Checked核查于 2026-10-15

Same models on other clouds同一模型，其他云平台价格

Some Chinese models are also sold through Alibaba Cloud Model Studio and BytePlus ModelArk, sometimes at different prices from the model maker.部分中国模型也可通过阿里云百炼、BytePlus 等平台调用，价格可能与厂商官方不同。

Alibaba Cloud Model Studio (third-party models, International scope) · DeepSeek / Moonshot / Zhipu

Same model families hosted by Alibaba Cloud, International deployment scope (Singapore). Prices can differ from the model maker's own API.阿里云百炼国际版托管的同系列模型（International 部署范围，新加坡）。价格可能与模型厂商官方 API 不同。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
deepseek-v4-pro$2.40Unknown未知$4.80Not stated价格页未注明
deepseek-v4-pro-0813$1.32Unknown未知$3.96Not stated价格页未注明Busy-hour price; idle hours $0.66 / $1.98 (busy = 08:00-22:00 UTC+8).
kimi-k3$3.00Unknown未知$15.00Not stated价格页未注明
glm-5.3$1.40Unknown未知$4.40Not stated价格页未注明

Source来源: https://www.alibabacloud.com/help/en/model-studio/model-pricing · Checked核查于 2026-10-15

BytePlus ModelArk (third-party models) · DeepSeek / Zhipu

Same model families hosted by BytePlus ModelArk, online inference (standard). BytePlus peak hours for deepseek-v4-1-flash: Mon-Fri 09:00-12:00 and 14:00-18:00 UTC+8.BytePlus ModelArk 托管的同系列模型（在线推理标准价）。deepseek-v4-1-flash 高峰时段：周一至周五 UTC+8 09:00-12:00、14:00-18:00。

Model模型Input输入Cached input缓存命中输入Output输出Context上下文Tier / notes计价档位 / 备注
deepseek-v4-1-flash-260910$0.30$0.006$1.20Not stated价格页未注明Peak price; off-peak $0.15 / $0.60.
deepseek-v4-pro-ga-260813$1.32$0.044$3.96Not stated价格页未注明
glm-5-2-260617$1.40$0.26$4.40Not stated价格页未注明
glm-5-3-flash-260828$0.15$0.03$0.50Not stated价格页未注明

Source来源: https://ai.byteplus.com/ark/region:ap-southeast-1/docs/model-pricing · Checked核查于 2026-10-15

Methodology方法说明