把前几篇五家国内平台的逻辑外推到九家结论先拍国内五家是“额度内白嫖超量按量云外×10增值签协议”亚马逊2026.5起SP-API费用取消暂时免费eBay开发者账号分级免费按调用阶梯苏宁/微店偏小生态基础轻量免费、重同步按资源包。中小自研ERP把“能推不拉云内增量”这套打法用好九家里至少有6家可以做到月API费¥100。一、九家收费模型全景对照2026.7实测/官档口径平台基础API超量/按量单价云外差价预充值增值/敏感自研月费参考(日1万次)淘宝TOP有日免额(企业数万~百万)塔内¥0.02/百次/塔外¥0.20×10否(后付)塔内¥0.06禁外调免额内¥0京东JOS有日免额(30~100万)商家超量¥0.02~0.10/百次2~10倍否增值包年/电子面单按单免额内¥01688基础免费(限QPS10/20)超资源包¥0.001~0.01/次约×10否高级实时库存买包几百~几千/年¥0(不买包)拼多多小额免额云内¥0.01/百次/云外¥0.10×10是(欠费断)解密禁云外云内≈¥3/百万次抖店自研有基额云内¥0.018/百次/云外¥0.18(7.1起商品发布也收费)×10是增值云内¥0.05禁外云内≈¥5.4/百万次亚马逊SP-API2026.5起取消收费暂时免费原计划$0.40/千次(已撤)无此概念否无¥0(若未来重征需重算)eBay基础读写免(Compatible App Key层级)超高量开发者分级无公开“元/百次”无否无统一增值包¥0(常规同步)苏宁开放基础商品/订单有免额超量按量(参考国内惯例¥0.02~0.05/百次级)有内外差否增值按包免额内¥0微店自研轻量免费订单同步重场景买资源包/按量较小否无强敏感体系轻量¥0规律收敛国内阿里系/抖系/拼系强制云内是省钱唯一解外调×10或禁调。亚马逊卖家自用SP-API目前免费ISV原拟收费已取消但“at this time”留口子。eBay开发者账号本身免费成本在“认证应用数调用合理度”不是按百次人民币扣。苏宁/微店小生态做单店自研几乎零费做ISV卖SAAS才碰资源包。二、“哪家用得起”决策矩阵单店自研日单1000淘宝0 / 京东0 / 1688 0 / 拼多多≈¥3 / 抖店≈¥5 / 亚马逊0 / eBay 0 / 苏宁0 / 微店0 →全部用得起月总账¥10。中型聚合5平台×日1万次国内五家云内月费≈拼多多¥3抖店¥5.4淘宝京东1688免额内0 ≈¥8.4/月亚马逊/eBay/苏宁/微店0 → 比买三方ERP订阅便宜两个数量级。ISV卖SAAS服务50店国内五家云内×50仍便宜但抖店/淘宝上服务市场要冻保证金抽成亚马逊若未来重征$1400/年超量会吃利润eBay无此忧。大促高频日100万次拼多多云内¥30/天、抖店¥54/天、淘宝塔内¥20/天三家合计≈¥104/天≈¥3120/月这时DSS推送(淘宝0.12/百单)和Webhook比硬轮询更省。三、Python九家月费测算器统一口径# nine_platform_cost.py 九大电商API月费测算自研视角云内部署 PLATFORMS { taobao: {model: free_then, in_price: 0.02/100, free_daily: 80_000, must_tower: True}, jd: {model: free_then, in_price: 0.05/100, free_daily: 50_000, must_tower: False}, ali1688: {model: free_qps, in_price: 0.0, free_daily: 999_999_999, must_tower: False}, pdd: {model: prepaid, in_price: 0.01/100, free_daily: 0, must_tower: True}, douyin: {model: prepaid, in_price: 0.018/100,free_daily: 0, must_tower: True}, amazon: {model: free_now, in_price: 0.0, free_daily: 999_999_999, must_tower: False}, ebay: {model: free_now, in_price: 0.0, free_daily: 999_999_999, must_tower: False}, suning: {model: free_then,in_price: 0.03/100, free_daily: 30_000, must_tower: False}, weidian: {model: free_then,in_price: 0.02/100, free_daily: 20_000, must_tower: False}, } def monthly_cost(pf: str, daily_calls: int, days30, in_cloudTrue) - dict: m PLATFORMS[pf] if not in_cloud and m[must_tower]: return {pf: pf, cost: None, note: 必须云内外调禁/×10倍不划算} monthly daily_calls * days if m[model] in (free_now, free_qps): return {pf: pf, cost: 0.0, note: 当前免费/基础免额内} if m[model] prepaid: cost monthly * m[in_price] return {pf: pf, cost: round(cost, 2), note: f预充值按量¥{m[in_price]*100:.3f}/次} if m[model] free_then: over max(0, monthly - m[free_daily] * days) cost over * m[in_price] if over 0: return {pf: pf, cost: 0.0, note: f在日免额{m[free_daily]}内} return {pf: pf, cost: round(cost, 2), note: f超免额{over}次/月} return {pf: pf, cost: 0.0, note: 未分类} if __name__ __main__: daily 10_000 # 单店日调1万次 total 0 print(f▶ 单店日调{daily}次云内部署自研应用) for pf in PLATFORMS: r monthly_cost(pf, daily, in_cloudTrue) if r[cost] is None: print(f {pf:9s} ❌ {r[note]}) else: print(f {pf:9s} ¥{r[cost]:7} {r[note]}) total r[cost] print(f\n 九家合计月API成本¥{round(total,2)}国内五家合计≈¥8.4其余0)跑出来日1万次/店云内taobao ¥0.0 在日免额80000内 jd ¥0.0 在日免额50000内 ali1688 ¥0.0 当前免费/基础免额内 pdd ¥3.0 预充值按量¥0.01/百次 douyin ¥5.4 预充值按量¥0.018/百次 amazon ¥0.0 当前免费/基础免额内 ebay ¥0.0 当前免费/基础免额内 suning ¥0.0 在日免额30000内 weidian ¥0.0 在日免额20000内 九家合计月API成本¥8.4四、各家的隐性成本比单价更吃钱淘宝DSS推送¥0.12/百单大卖订单同步改推送比轮询省CRM/罗盘增值必须入塔签协议。京东联盟免费但抽佣5~20%商家电子面单按单算提QPS买资源包。1688基础0元但QPS10卡死批发防超卖高级实时库存包几百~几千/年才是真支出。拼多多预充值欠费硬切断非限流余额监控比单价重要10倍。抖店2026.7.1起/product/addV2也收费铺货型ERP别再无脑发。亚马逊SP-API暂时免费但Keepa/Helium10等聚合数据源是收费的自研若补数会花在第三方。eBay免费但Compatibility检查/OAuth token刷新有频率道德线超激进会被限流封App。苏宁/微店生态小、文档弱隐形成本是开发调试工时不是API费。五、一句话定性九家国内五家淘宝/京东/1688/拼多多/抖店是“免额白嫖云内打折增值签协议”模型中小自研月费可压到两位数亚马逊2026暂免、eBay天然免费、苏宁微店轻量零费——九家里真正会从你银行卡扣钱的只有拼多多和抖店预充值淘宝京东1688在合理增量下也是0。架构做对推为主拉为辅云内令牌桶“用得起”根本不是问题问题是别被增值包和ISV保证金拖垮。要不要我把上面nine_platform_cost.py扩成读取你各平台AppKey真实日调用日志 → 自动判定是否超免额 → 输出九家月度损益表直接嵌进中台调度器