Juga LabsOne API key. Every frontier model.
Stop juggling keys, dashboards and invoices across eight providers. One endpoint speaks OpenAI, Anthropic and Gemini natively, so the SDK and CLI tools you already run keep working.
https://models.zuxoai.cn/v1
Point any OpenAI SDK, Codex, Cline, Cursor or desktop client at this base URL.
Authorization: Bearer sk-...
https://models.zuxoai.cn
Claude Code and the Anthropic SDK talk to /v1/messages with no translation layer.
x-api-key: sk-...
https://models.zuxoai.cn
Gemini CLI and google-genai call /v1beta directly, including streaming.
x-goog-api-key: sk-...
Two lines of config, then nothing changes.
Works with the OpenAI SDKs, the Anthropic SDK, google-genai, Claude Code, Codex, Gemini CLI, OpenClaw, Cline, Continue, Cursor and Cherry Studio.
from openai import OpenAI
client = OpenAI(
base_url="https://models.zuxoai.cn/v1",
api_key="sk-...",
)
resp = client.chat.completions.create(
model="claude-opus-5",
messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)One key does all of it.
Not on the shared endpoint today: embeddings, speech to text, content moderation.
Routing runs on infrastructure operated with our partner ZUXO.
Prepaid USD. No subscription.
Retail rates, published in full. Every model is at least 20 percent below OpenRouter for the same model, and above ten thousand dollars a month we price against your actual traffic.
| Model | Context | Input / 1M | Output / 1M | Vs OpenRouter |
|---|---|---|---|---|
Claude Opus 5 Anthropic | 1000K | $2.250 | $11.25 | 55% less |
Claude Sonnet 5 Anthropic | 1000K | $1.350 | $6.750 | 33% less |
Claude Haiku 4.5 Anthropic | 200K | $0.420 | $2.140 | 57% less |
GPT 6 Astra OpenAI | 1050K | $4.280 | $25.71 | 50% less |
GPT 5.6 Sol OpenAI | 1050K | $1.600 | $8.000 | 20% less |
GPT 5.6 Terra OpenAI | 1050K | $0.900 | $5.410 | 55% less |
GPT 5.6 Luna OpenAI | 1050K | $0.160 | $0.960 | 20% less |
GPT 5.5 OpenAI | 128K | $2.140 | $12.85 | 57% less |
Gemini 3.1 Pro Google | 1000K | $0.900 | $5.410 | 55% less |
Gemini 3.8 Flash Google | 1000K | $0.600 | $3.000 | 20% less |
from $0.03 per image
GPT Image 2 and the Nano Banana family
from $0.14 per second
Seedance, Kling, Wan and Hailuo, async job API
$0.43 per 10k characters
327 preset voices plus voice cloning
90 percent off input
Charged at a tenth of the input rate
What your current workload would cost.
Estimate only. Cached input is billed at a tenth of the input rate, so agent and coding workloads usually land below this.
Top up any amount. Spend it when you like.
Credit sits on your account, never expires into a subscription, and larger top ups carry bonus credit.
$20
Test the endpoint against your own workload.
$100plus 5%
For a team running agents and coding clients daily.
$500plus 10%
Production traffic with predictable spend.
Above $10,000 a month
Published rates are retail. At this level we price against your actual model mix and traffic pattern, so talk to us before you top up.
- Negotiated per model rates
- Invoiced billing terms
- Dedicated routing capacity
- Direct line to the team
A funded account, then your keys.
Keys are issued against a funded account, never on their own. That is what lets us give each key its own credit limit and show you exactly where the balance went.
Create your account
Send your details and the models you plan to use. An account is opened in your name, one per company or developer.
Fund it
Any amount from ten dollars up, by crypto or card. An account has to be funded before it can hold a key, so there are no idle keys and no free tier to abuse.
Get your keys
Each key belongs to your account and carries its own credit limit, so a staging key cannot spend the production budget. Issue as many as you need.
Watch the spend
Usage is tracked per key and per model against your balance. Top up the same account at any time without touching your integration.
- Self serve signup and card top ups from inside your account
- An API that generates keys and sets their credit limits programmatically
- Per key spend alerts and monthly statements
Against the alternatives.
| AI Router | OpenRouter | Vendor accounts | |
|---|---|---|---|
| Price per model | At least 20 percent lower | Public list | Public list |
| Keys to manage | One | One | One per vendor |
| Protocols | OpenAI, Anthropic and Gemini native | OpenAI compatible | Each vendor, separately |
| Top up | Any amount, crypto or card | Card or crypto | Card per vendor |
| Volume pricing | Negotiated above $10k a month | Published rates | Enterprise agreement |
| Support | Direct line to our team | Docs and community | Vendor support tiers |
We buy model capacity in bulk through our infrastructure partner and pass most of that discount to you. The same upstream providers serve your requests, so quality does not change with the price.
Before you switch.
No. One key covers every model on the list, across all three protocols. Switch models by changing the model string in your request, nothing else.
Create an account and fund it. Keys are issued against a funded account only, which is how spend stays tied to a real balance. Tell us your volume, pay by crypto or card, and your first key is issued with the credit limit you choose.
Yes. Keys belong to your account and each one carries its own credit limit, so you can cap a staging key or a client project without exposing the whole balance.
USDT or USDC on major chains, BTC, or a card through our fiat payment gateway. Credit lands on the key once the payment confirms.
Every published rate sits at least 20 percent below OpenRouter for the same model, and on several models the gap is much wider. Above ten thousand dollars a month we price against your actual traffic, so the gap usually widens again.
Only the base URL and the key. The endpoint speaks the OpenAI, Anthropic and Gemini protocols, so existing SDKs and CLI tools work as they are.
Calls return a 403 with insufficient quota and nothing else changes. Top up and the same key keeps working. There is no subscription and no auto charge.
Not on the shared endpoint today. Small embedding models run well locally, and we can set that up as part of an advisory engagement.
Send us your workload.
Tell us the models and the monthly volume. You get a rate, a key and a test budget, so you can benchmark it against what you pay now.