# NoviaHub

> NoviaHub is an AI model access platform operated by NexaMind Singapore Pte. Ltd. One account, one balance and one API key give access to chat and image models from OpenAI (GPT), Anthropic (Claude), Google (Gemini), DeepSeek, Moonshot (Kimi) and Zhipu (GLM), through the OpenAI, Anthropic and Google Gemini API formats. Existing official SDKs and AI coding tools work by changing the base URL and key. Website: https://noviahub.com · Docs: https://docs.noviahub.com/en/ (English) and https://docs.noviahub.com (Chinese)

## Base URL & auth
- Base URL: https://noviahub.com
  - OpenAI SDKs: `https://noviahub.com/v1`
  - Anthropic SDKs: `https://noviahub.com` (the SDK adds `/v1/messages`)
  - Google Gen AI SDKs: `https://noviahub.com` (the SDK adds `/v1beta/models/...`)
- Auth header: `Authorization: Bearer <API_KEY>` works on every endpoint. `x-api-key` is also accepted on `/v1/messages` and `/v1/models`, and `x-goog-api-key` on the Gemini model paths.
- API keys start with `sk-`. Sign up at https://noviahub.com and create a key in the console.
- Every response carries a request ID (header `X-Novia-Request-Id`; error messages end with `(request id: ...)`). Include it when reporting a failed call.

## Quickstart (Python, OpenAI SDK)
```
from openai import OpenAI
client = OpenAI(api_key="<API_KEY>", base_url="https://noviahub.com/v1")
resp = client.chat.completions.create(
    model="<MODEL_ID>",  # any model ID listed on https://noviahub.com/pricing
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)
```

## Endpoints
- OpenAI: `POST /v1/chat/completions`, `POST /v1/responses`, `POST /v1/images/generations`, `GET /v1/models`
- Anthropic: `POST /v1/messages`, `GET /v1/models`
- Gemini: `POST /v1beta/models/{model}:generateContent`, `POST /v1beta/models/{model}:streamGenerateContent`, `GET /v1beta/models`
- The endpoint labels of each model on https://noviahub.com/pricing show which formats it accepts; see https://docs.noviahub.com/en/api/protocol-conversion/ for how formats are converted.

## Pricing & billing
- Charged for actual usage from your account balance, shown in US dollars. Top up in the console wallet.
- Every model's price is public at https://noviahub.com/pricing: input, output, cache read and cache write per 1M tokens, or per image for image models. Some models use length tiers or time-of-day prices; the model's page is authoritative.
- The usage and cost of every call are recorded in the console's usage logs.
- Details: https://docs.noviahub.com/en/billing/overview/

## Key pages
- Models & pricing: https://noviahub.com/pricing (one model: https://noviahub.com/pricing/<MODEL_ID>)
- Benchmarks (capability index, LMArena ratings and other public leaderboards per model): https://noviahub.com/benchmarks/models
- Benchmark tests: https://noviahub.com/benchmarks/tests
- Degradation radar (public signals on whether a model got worse at a provider): https://noviahub.com/benchmarks/radar
- Rankings (model and app usage on NoviaHub): https://noviahub.com/rankings/models and https://noviahub.com/rankings/apps
- About: https://noviahub.com/about
- Sitemap: https://noviahub.com/sitemap
- Developers (AI news, open-source AI rankings, Agent Skills, MCP servers, prompts and templates): https://noviahub.com/developers (English) and https://noviahub.com/zh/developers (Chinese); sections: AI News https://noviahub.com/developers/news, Open Source Trending https://noviahub.com/developers/repos, Skills https://noviahub.com/developers/skills, MCP https://noviahub.com/developers/mcp, Prompts https://noviahub.com/developers/prompts, Templates https://noviahub.com/developers/templates

## Docs
- Quickstart: https://docs.noviahub.com/en/getting-started/quickstart/
- API overview: https://docs.noviahub.com/en/api/overview/
- Coding tools (Claude Code, Codex, Cursor, Cline, Roo Code, Kilo Code, OpenCode, Zed, Qwen Code, Kimi Code, Cherry Studio, cc-switch and more): https://docs.noviahub.com/en/integrations/overview/
- Guides: streaming https://docs.noviahub.com/en/guides/streaming/ · tool calling https://docs.noviahub.com/en/guides/tool-calling/ · reasoning https://docs.noviahub.com/en/guides/reasoning/ · vision https://docs.noviahub.com/en/guides/vision/ · prompt caching https://docs.noviahub.com/en/guides/prompt-caching/ · image generation https://docs.noviahub.com/en/guides/image-generation/
- Errors and FAQ: https://docs.noviahub.com/en/help/errors/ and https://docs.noviahub.com/en/help/faq/

## Privacy & legal
- NoviaHub does not use request content to train models; see https://noviahub.com/about
- Privacy Policy: https://noviahub.com/privacy-policy
- Security & Compliance: https://noviahub.com/security-policy
- Terms of Service: https://noviahub.com/user-agreement

## Contact
- Support: support@noviahub.com
- Discord: https://discord.gg/zQ3HRH9sz6