Access frontier Chinese models like DeepSeek through a single OpenAI-compatible endpoint — also compatible with Claude Code and Anthropic SDK. Pay-as-you-go with USDT or PayPal (TRC20). No Chinese phone number, no VPN, no hassle.
We call DeepSeek's official API directly, on pay-as-you-go billing, with multi-path upstream routing and automatic failover. No expiry date on our upstream access, and no shared or one-off key pool. We are not the cheapest — the cheap ones are cheap because they are temporary.
Frontier-level capability at a fraction of the cost.
Drop-in replacement for the OpenAI SDK. Change the base URL and API key — your existing code just works.
No Chinese phone number or ID verification needed. Sign up with just an email and start building from anywhere.
Pay-as-you-go with USDT or PayPal (TRC20), from $5. No subscription required, no wasted tokens — you only pay for what you use.
Card declined, or no PayPal where you are? How to pay for a DeepSeek API key in blocked countries walks through the four routes that actually work, country by country.
We call DeepSeek's official API directly, on pay-as-you-go billing, with multi-path upstream routing and automatic failover. No expiry date on our upstream access, and no shared or one-off key pool. We are not the cheapest — the cheap ones are cheap because they are temporary.
Top-tier models like DeepSeek V4.1 deliver strong reasoning at a fraction of comparable Western API prices.
One endpoint wired straight to DeepSeek's official API — no aggregation layer and no shared account pool. There is no SLA: we publish no availability percentage rather than a figure we have not measured.
Drop-in for Claude Code, Codex, Cursor, Cline, Cherry Studio, Dify and any tool that speaks the OpenAI or Anthropic API format.
Same tools, same workflows — just point your base URL at NovaAPI and use a Chinese frontier model.
Per-token pricing. Pay as you go — no subscription, no hidden fees.
We call DeepSeek's official API directly, on pay-as-you-go billing, with multi-path upstream routing and automatic failover. No expiry date on our upstream access, and no shared or one-off key pool. We are not the cheapest — the cheap ones are cheap because they are temporary.
| Model | Best for | Usage cost |
|---|---|---|
| DeepSeek V4 Flash | Fast daily driver for coding agents | about 70–85% lower |
| DeepSeek V4 Pro | Complex reasoning, deep code work | lower again |
| DeepSeek V4 Flash Vision | Chat with image input | about 70–85% lower |
"Usage cost" compares the same monthly workload on our endpoint against the same workload priced at Claude Sonnet's published list rates. The range depends on your mix of input and output tokens, so we quote a range rather than a point estimate. How the comparison is built.
Model names stay deepseek-flash / deepseek-v4-pro in the API — requests are now served by the latest DeepSeek V4.1 Flash engine.
Every Claude model name your tool already sends — Sonnet 5, Sonnet 4.6, Haiku 4.5, Opus 5 / 4.6 / 4.5, Fable 5 / 5.1 — is routed for you. In most clients you change the base URL and the key, and nothing else.
Usage cost: about 70–85% lower than the same workload priced at Claude Sonnet's published list rates.
Claude-compatible names keep your existing Claude Code / Anthropic SDK workflow untouched, at roughly 70–85% below Anthropic’s list prices.
If you have used the OpenAI API, you already know how to use NovaAPI.
Create an account with your email and $1 of free credit lands in your balance instantly — no card, no payment needed to start. When you are ready, top up from $5 with USDT or PayPal.
We call DeepSeek's official API directly, on pay-as-you-go billing, with multi-path upstream routing and automatic failover. No expiry date on our upstream access, and no shared or one-off key pool. We are not the cheapest — the cheap ones are cheap because they are temporary.
Use this base URL everywhere a tool asks for an API address:
https://api.lbase.com/v1
OpenAI-compatible. Claude Code & Anthropic SDK point at https://api.lbase.com.
Pick a model and go. Works with the OpenAI SDK, LangChain, Cursor, Claude Code and any OpenAI-compatible tool.
No. Every new account is credited with $1 of free credit the moment you sign up — create an API key and send real requests straight away. When you want more, the minimum top-up is $5 via USDT or PayPal.
No. We designed NovaAPI for global developers — sign up with just an email address and pay with USDT or PayPal.
We currently offer DeepSeek V4 models (native) plus Claude-compatible endpoints backed by our high-performance Chinese model network. More Chinese model families (Qwen, GLM, Kimi and MiniMax) are being onboarded and will appear in the console as they go live.
Yes. We expose an OpenAI-compatible API. Point your OpenAI SDK or tools such as LangChain, Dify or Cursor at our endpoint and you are ready.
Pay-as-you-go. You top up your account balance via USDT or PayPal and are charged per token used. No monthly subscription.
Requests are routed to the model providers you use. See our Privacy Policy below for details. Requests are answered by the model provider you select, and screened by an automated moderation check before they are forwarded. The Privacy Policy sets out exactly what is stored, what is not, and for how long.
Last updated: September 2026
Effective date: September 1, 2026
By creating an account or using the NovaAPI service (“the Service”), you agree to these Terms of Service. If you do not agree, please do not use the Service.
NovaAPI provides a unified application programming interface (API) that routes requests to third-party large language model providers. The Service currently offers DeepSeek V4 models as well as Claude-compatible endpoints backed by high-performance Chinese model infrastructure; additional Chinese model families such as Qwen, GLM, Kimi and MiniMax are added as they become available. We act as an access layer; the underlying models are operated by their respective providers.
The Service is provided “as is” without warranties of any kind. Upstream model providers may change, deprecate, or limit models at any time. To the maximum extent permitted by law, we are not liable for indirect, incidental, or consequential damages, or for any loss arising from your use of the Service or the output of third-party models.
Either party may terminate this agreement. Upon termination you must stop using the Service; any remaining balance will be handled per our refund policy.
These terms are governed by the laws of Hong Kong SAR. Any disputes shall be resolved in the courts of Hong Kong SAR.
Effective date: September 1, 2026
To fulfill your requests, prompts and related data are forwarded to the third-party model provider you select. Those providers process data under their own privacy policies. We do not sell your personal data.
Request logs are retained for a limited period needed for billing, abuse prevention and debugging, after which they are deleted or anonymized.
We use industry-standard measures to protect account credentials and data in transit. No method of transmission is 100% secure; you should not submit highly sensitive information through the Service.
You may request access to, correction of, or deletion of your personal data by contacting us. We will respond within a reasonable timeframe.
For privacy or legal inquiries: service@lbase.com