Prepaid AI API relay预付费 AI API 中转
Claude, GPT, Grok, GLM, Kimi and Jev. One base URL, one key. Claude、GPT、Grok、GLM、Kimi、Jev。一个 Base URL，一把密钥。
TopxAI relays thirteen models from six providers through OpenAI- and Anthropic-compatible endpoints. Every model has a fixed USD price on every route, visible before you create a key. Prompts and completions pass through memory and are never written to disk. TopxAI 把六家的十三个模型接到兼容 OpenAI 和 Anthropic 协议的端点上。每个模型在每条线路上的美元价格都是固定的，建密钥之前就能看到。提示词和回复只在内存里经过，不写磁盘。
- Base URL
https://ai.topxea.com/v1- Console控制台
- ai.topxea.com
- Protocols协议
- OpenAI Chat, Responses, Images, Video; Anthropic Messages; System OneOpenAI Chat、Responses、Images、Video；Anthropic Messages；System One
- Billing计费
- Prepaid USD. Per token, per image, per second.预付美元。按 token、按张、按秒。
- Retention留存
- Zero. Logs hold counts, models, routes and ids.零。日志只留 token 数、模型名、线路和标识。
Keep your SDK. Change two lines.SDK 不用换，改两行。
Point the base URL at TopxAI and use a TopxAI key. The OpenAI and Anthropic SDKs, LiteLLM, Cursor, Claude Code and anything else that speaks either protocol keep working as they are.Base URL 指到 TopxAI，换上 TopxAI 的密钥。OpenAI SDK、Anthropic SDK、LiteLLM、Cursor、Claude Code，以及其他任何按这两个协议说话的工具，原样能用。
from openai import OpenAI
client = OpenAI(
base_url="https://ai.topxea.com/v1",
api_key="sk-...", # your TopxAI key
)
r = client.chat.completions.create(
model="gpt-6-astra",
messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)import anthropic
client = anthropic.Anthropic(
base_url="https://ai.topxea.com",
api_key="sk-...", # your TopxAI key
)
msg = client.messages.create(
model="claude-fable-5-1",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello"}],
)
print(msg.content[0].text)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://ai.topxea.com/v1",
apiKey: process.env.TOPXAI_KEY,
});
const r = await client.chat.completions.create({
model: "grok-4.6",
messages: [{ role: "user", content: "Hello" }],
});
console.log(r.choices[0].message.content);curl https://ai.topxea.com/v1/chat/completions \
-H "Authorization: Bearer $TOPXAI_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"claude-sonnet-5","messages":[{"role":"user","content":"Hello"}]}'Four things the console does not hide.控制台不藏着的四件事。
-
Explicit model choice模型选择清晰
Choose your model and route. Know the price before the first request.模型与线路由你选择，发起请求前即可了解价格。
-
One connection, six providers一次接入，六家模型
Use familiar OpenAI and Anthropic APIs with your existing tools.沿用现有工具，使用熟悉的 OpenAI 与 Anthropic API。
-
Zero retention零留存
Prompts and completions are relayed, never stored.提示词和回复只经转发，从不存储。
-
Clear usage-based pricing按量计费，价格透明
Compare model prices and follow every charge in your usage log.逐项对比模型价格，在用量日志中查看每一笔费用。
13 models, 6 lines, fixed USD prices.13 个模型，6 条线路，固定美元价。
Same upstream, different routing and price. The shared pool is 50% of list, the official line 90%. Kimi is 80%, Jev 107%. GLM runs on a private deployment above list. Text is billed per million tokens, images per picture, video per second.上游相同，路由方式和价格不同：共享池按牌价 50% 计，官方线路 90%。Kimi 为 80%，Jev 为 107%。GLM 跑在私有部署上，高于牌价。文本按每百万 token 计，图片按张，视频按秒。
Seven paths under one host.一个域名下的七条路径。
Every path lives under ai.topxea.com and accepts the same TopxAI key. Which models each path serves is listed on the cards above.全部路径都在 ai.topxea.com 下，用同一把 TopxAI 密钥。每条路径服务哪些模型，上面的卡片里有标。
| Protocol协议 | Path路径 | Serves用途 |
|---|---|---|
| OpenAI Chat | POST /v1/chat/completions | Claude, GPT, Grok, GLM and Kimi text models; image models in chat formClaude、GPT、Grok、GLM、Kimi 文本模型；图片模型的对话形式 |
| OpenAI Responses | POST /v1/responses | GPT-6 Astra, GPT-5.6 Sol, Grok 4.6GPT-6 Astra、GPT-5.6 Sol、Grok 4.6 |
| Anthropic Messages | POST /v1/messages | Claude Fable 5.1, Fable 5, Sonnet 5, Opus 5Claude Fable 5.1、Fable 5、Sonnet 5、Opus 5 |
| Images | POST /v1/images/generations | GPT Image 2.5 Sunburst and Flare, priced per pictureGPT Image 2.5 Sunburst 与 Flare，按张计价 |
| Image edit | POST /v1/images/edits | The same two image models, with an input picture同样两个图片模型，带输入图 |
| Video | POST /v1/videos/generations | Grok Imagine Video 1.5, priced per second by resolutionGrok Imagine Video 1.5，按分辨率每秒计价 |
| System One | POST /v1/systemone | TypeSafe Jev: typed judgments with calibrated probabilities, input tokens onlyTypeSafe Jev：带校准概率的类型化判断，只计输入 token |
Prepaid USD. Spent only by your own requests.预付美元，只被你自己的请求花掉。
The wallet balance is the source of truth. Each request holds its cost, settles once, and writes one usage-log row with the model, the route and the charge. Historical bills are never recomputed from today's prices.钱包余额是唯一依据。每个请求先冻结费用，结算一次，在用量日志里写一行：模型、线路、扣费。历史账单不会按今天的价格重算。
Card and local methods银行卡与本地支付
Completed on topxea.com through its checkout and settled back to your TopxAI wallet. One sign-in covers both sites.在 topxea.com 的收银台完成，再结算回 TopxAI 钱包。两个站共用一套登录。
Crypto invoice加密货币账单
Hosted by NOWPayments. Pick the asset and network in the checkout; the wallet is credited after server-side confirmation, never from a return URL.由 NOWPayments 托管。在收银台选币种和网络；服务端确认后才入账，不认返回链接。
Single-use credit codes一次性储值码
Fixed amount and expiry. Redeeming twice is safe: the second attempt returns the original receipt without a second credit.金额和有效期固定。重复兑换没事：第二次只返回原始凭证，不会二次入账。
Before you create a key.建密钥之前。
Do I need a separate key for each model?每个模型都要单独建一个密钥吗？
No. One auto-route key covers every model and picks the lowest-priced line for each request. Per-route keys are there for when you want to pin a line.不用。一个自动路由密钥覆盖全部模型，每个请求都走价格最低的线路。想固定走某条线路时，再建按线路的密钥。
Which price applies to a request?一次请求按什么价格计费？
The fixed USD price of the model on the route the key uses, as listed on the models table before you create the key. The usage log shows the actual charge afterwards.按密钥所用线路上该模型的固定美元价格，建密钥之前就能在模型表里看到。请求之后，用量日志会显示实际扣费。
Is my data stored or used for training?我的数据会保存下来或拿去训练吗？
No. Content passes through in memory and is not written anywhere. What TopxAI keeps is described in the Privacy Policy. Upstream providers receive the content you send and apply their own data policies.不会。内容只在内存中经过，不写到任何地方。TopxAI 保留哪些信息，隐私政策里有说明。上游厂商会收到你发出的内容，并按各自的数据政策处理。
What is GLM-5.3-Abliterated?GLM-5.3-Abliterated 是什么？
A fine-tuned GLM 5.3 with its refusal behaviour removed, served from a private deployment rather than the vendor API. Text only, priced above list, and covered by the same Acceptable Use Policy as every other line.在 GLM 5.3 基础上微调、去掉了拒答行为的模型，运行在私有部署上，而不是厂商 API。只支持文本，价格高于牌价，和其他线路一样受可接受使用政策约束。
Where do I sign in?在哪里登录？
The console is at ai.topxea.com. It shares sign-in with topxea.com: if you already have a session there, the console picks it up. This page is an introduction only.控制台在 ai.topxea.com，和 topxea.com 共用登录：那边已经登录的话，控制台会直接接上。这个页面只做介绍。
TopxAI does not provide any services to users located in mainland China. TopxAI 不对中国大陆地区用户提供任何服务。