《实测：一家店跑九家API一个月花多少？》（附Python源码）

先纠标题两个误读第一亚马逊1400∗∗不是单店账单∗∗——那是2025.11原拟方案里∗∗第三方ISV开发者年费∗∗0.40/千次GET超量也是针对ISV且这套结构已在2026-05-12官方正式取消will not move forward at this time当前卖家自用/第三方ISV调用费均为$0 。第二实测结论可能让你意外——九家API调用费本身极低单店月API费普遍在¥0~¥41区间真正花钱的是云资源ECS1688资源包而非按量扣费。一、实测结论三种典型单店画像画像接入平台月调用量API月费资源包摊销ECS(估)月度OPEX​自研标品单店淘宝京东1688拼多多抖店(云内)~6.6万¥3.67​¥0¥250¥253.67​ISV单店(服1商家)九家全接(云内)~51万¥37.34​¥165(1688包)¥250¥452.34​铺货型单店5家跨铺自动上新~35.4万¥41.34​¥165(1688包)¥250¥456.34​关键洞察API按量费最高才¥41/月铺货型最低¥3.67/月自研标品。淘宝/京东/苏宁因为企业日免额内(8万/5万/1万次)调用费¥01688基础免费亚马逊$0。真正让ISV/铺货型多花¥200的是1688高级资源包¥1980/年÷12≈¥165/月——这是为提频(QPS 10→50/100/200)和实时库存付费不是按次扣 。二、九家逐家拆解云内部署口径2026.08现行平台计费模式云内单价月费(自研/ISV/铺货)一句话定性淘宝TOP​免额内0超量按量基础¥0.02/百次增值¥0.06¥0/¥2/¥0企业日免额8万自研几乎不超京东JOS​免额内0超量按量¥0.05/百次(云外×3)¥0/¥5/¥0面单按单另计1688​基础免费资源包年费¥0(QPS10)包¥980~2980/年¥0/¥165/¥165不按次扣提频才买包拼多多​预充值按量基础¥0.01/百次增值¥0.03¥1.5/¥9/¥7.5欠费硬断(非限流)云外×10抖店​预充值按量基础¥0.018/百次增值¥0.05¥2.17/¥16.24/¥22.142026.7.1起商品发布收费苏宁​免额内0超量~¥0.03/百次(云外×3)¥0/¥1.5/¥0小生态微店​轻量自研免费¥0(QPS5)¥0/¥0/¥0重同步按需买包快手​参考抖系~¥0.018/百次(云外×10)¥0/¥3.6/¥11.7无独立沙箱亚马逊SP-API​当前$0(原拟撤)$00/0/$0Basic 2.5M GET警戒线留后手拼多多/抖店是九家里唯一真正按量扣钱且单价不为零的两家¥0.01/¥0.018但单价极低——铺货型月调用12万次的抖店也才¥22。真正的成本杀手是云外部署×10和欠费硬断不是单价本身 。三、亚马逊$1400的真相标题误读澄清原拟1400/年是∗∗第三方ISV开发者年费∗∗非卖家Basic档含2.5MGET/月超量0.40/千次卖家自用Private App历来免费、现在仍免费​ 。实测同主体多店敞口若原拟复活同主体店数月GET总量超量GET超量费/月含年费分摊/月1店19,5000$0$116.675店97,5000$0$116.6720店390,0000$0$116.67100店1,950,0000$0$116.67同主体20店以内月GET都压在Basic 2.5M免额内**超量费0∗∗仅年费分摊116.67/月。100店(分8个Key)才开始触碰超量线。这解释了为什么$1400对单店是噪声——它根本不是单店账单项 。四、Python九家单店月度成本实测测算器# single_shop_9platform_monthly.py 九家电商API 单店月度成本实测测算器 (2026-08口径) - 三种典型画像: 自研标品 / ISV单店 / 铺货型 - 分别输出九家月调用量/单价/API费/资源包摊销 - 亚马逊单独列若原拟复活敞口对照 计费参数来源: 各平台官方《开放平台技术服务费规则》(2026现行) from dataclasses import dataclass from typing import Dict # 封装好API供应商demo urlhttps://console.open.onebound.cn/console/?iLex PLAT { taobao: {kind:A,label:淘宝TOP,free:80000,in:0.02/100,out:0.20/100,val_in:0.06/100}, jd: {kind:A,label:京东JOS,free:50000,in:0.05/100,out:0.15/100}, 1688: {kind:B,label:1688,qps:10,pack_year:(980,2980)}, pdd: {kind:A,label:拼多多,free:0,in:0.01/100,out:0.10/100,val_in:0.03/100}, douyin: {kind:A,label:抖店,free:0,in:0.018/100,out:0.18/100,val_in:0.05/100}, suning: {kind:A,label:苏宁,free:10000,in:0.03/100,out:0.10/100}, weidian: {kind:B,label:微店,qps:5}, kuaishou:{kind:A,label:快手,free:10000,in:0.018/100,out:0.18/100}, amazon: {kind:C,label:亚马逊SP-API,free:None}, } dataclass class ShopProfile: name: str; desc: str; in_cloud: bool daily: Dict[str,int] value: Dict[str,bool] publish_month: Dict[str,int] buy_pack: Dict[str,bool] def unit(plat:str, in_cloud:bool, value:bool)-float: pPLAT[plat] if p[kind]!A: return 0.0 if value: return p[val_in] if in_cloud else 0.30/100 return p[in] if in_cloud else p[out] def monthly_cost(plat:str, prof:ShopProfile)-Dict: pPLAT[plat]; dprof.daily.get(plat,0); monthlyd*30 if p[kind]C: return {calls:monthly,api_cost:0.0,note:2026.05.12取消,调用费$0;Basic 2.5M GET警戒线} if p[kind]B: if plat1688: packround(1980/12,2) if prof.buy_pack.get(plat) else 0.0 return {calls:monthly,api_cost:0.0,pack_amort:pack, note:基础免费(QPS10)(f,资源包¥1980/年摊¥{pack}/月 if pack else ,无资源包)} if platweidian: return {calls:monthly,api_cost:0.0,pack_amort:0.0,note:基础免费(QPS5),重同步按需买包} freep[free]; overmax(0,monthly-free) costover*unit(plat,prof.in_cloud,prof.value.get(plat,False)) pubprof.publish_month.get(plat,0); pub_cost0.0 if platdouyin and pub0: pub_costpub/100 * 0.018 # 7.1起商品发布收费 return {calls:monthly,api_cost:round(costpub_cost,4), over_calls:over,unit:f¥{unit(plat,prof.in_cloud,prof.value.get(plat,False))*100:.4f}/百次, publish:pub,publish_cost:round(pub_cost,4),note:} def amazon_proposed_exposure(sellers:int,get_per_seller:int19500,keys:int1)-Dict: totalsellers*get_per_seller; pertotal/keys overmax(0,per-2_500_000)*keys; over_feeover/1000 * 0.40 return {sellers:sellers,keys:keys,get_month:f{total:,},over:f{int(over):,}, over_fee:round(over_fee,2),monthly:round(1400/12*keysover_fee,2)} def profile_ziying(): return ShopProfile(自研标品,5家国内自营,云内,推送替轮询,True, {taobao:400,jd:300,1688:500,pdd:500,douyin:400},{},{douyin:30},{1688:False}) def profile_isv(): return ShopProfile(ISV单店,服务商帮1商家接九家,云内,True, {taobao:3000,jd:2000,1688:2000,pdd:3000,douyin:3000, suning:500,weidian:500,kuaishou:1000,amazon:2000}, {taobao:False,pdd:False,douyin:False},{douyin:200},{1688:True}) def profile_puhuo(): return ShopProfile(铺货型,跨5家铺货,自动上新量大,True, {taobao:1500,jd:800,pdd:2500,douyin:4000,kuaishou:2500,suning:300,weidian:200}, {},{douyin:3000,kuaishou:1500},{1688:True}) def print_report(prof:ShopProfile): print(f\n{*70}\n画像: {prof.name} ({prof.desc})\n{*70}) print(f{平台:9}{月调用:9}{单价/百次:12}{API月费:11}{资源包/月:11}{备注:24}) print(-*70) t_apit_pack0.0 for plat in [taobao,jd,1688,pdd,douyin,suning,weidian,kuaishou,amazon]: rmonthly_cost(plat,prof); packr.get(pack_amort,0.0) t_apir[api_cost]; t_packpack print(f{PLAT[plat][label]:9}{r[calls]:9,}{r.get(unit,—):12}{¥format(r[api_cost],.2f):11}{¥format(pack,.2f):11}{(r[note][:23] if r[note] else (发布str(r.get(publish,0))次) if r.get(publish,0) else ):24}) totalt_apit_pack print(-*70) print(f{合计:9}{:9}{:12}{¥format(t_api,.2f):11}{¥format(t_pack,.2f):11} API月度总成本 ¥{total:.2f}) print(f 云资源(ECS共享,估): ¥250/月) print(f ★ 月度真实OPEX ≈ ¥{total250:.2f} (API¥{t_api:.2f}包¥{t_pack:.2f}ECS¥250)) if __name____main__: for fn in (profile_ziying,profile_isv,profile_puhuo): print_report(fn()) print(f\n{*70}\n亚马逊SP-API 敞口防守(若原拟复活,同主体多店)\n{*70}) for n in (1,5,20,100): eamazon_proposed_exposure(n) print(f {e[sellers]:4}店 {e[keys]}Key | GET/月{e[get_month]:12} 超量{e[over]:10} 超量费${e[over_fee]:8} 月合计${e[monthly]})跑出来的核心三行与上文报表一致自研标品 → API月费¥3.67 ECS¥250 月度OPEX ¥253.67 ISV单店 → API¥37.34 1688包¥165 ECS¥250 月度OPEX ¥452.34 铺货型 → API¥41.34 1688包¥165 ECS¥250 月度OPEX ¥456.34五、给老板的一句话成本定性九家API按量费不是成本问题是几乎免费——自研标品单店月API费¥3.67ISV/铺货型也才¥37~¥41。真正决定账单的是三件事① 云资源ECS¥250/月占OPEX 60%② 1688高级资源包¥165/月占ISV/铺货型 36%③ 千万别云外部署——拼多多/抖店/淘宝/快手云外×10一旦踩中单店月费能从¥41跳到¥400​ 。至于亚马逊1400——对单店是伪命题自用PrivateApp历来免费ISV年费已取消同主体20店内都落在Basic免额内超量费0。把预算从担心API按量费挪到云内着色推送替轮询1688包按需才是单店控本的正解。源码文件如下可直接python3 single_shop_9platform_monthly.py跑出你自己场景的九家月度账单single_shop_9platform_monthly.py