feat: add Novita AI as LLM provider

[Alex-wuhu]

Summary

• Registers Novita AI models in PROVIDER_MODELS and PROVIDER_MODELS_PREFIXES (crawl4ai/config.py) so NOVITA_API_KEY is auto-resolved for any novita/… provider string
• Documents NOVITA_API_KEY and NOVITA_BASE_URL in the Docker deployment README

Usage

Novita AI exposes an OpenAI-compatible endpoint. Use it with LLMConfig:

from crawl4ai import LLMConfig

llm_config = LLMConfig(
    provider="novita/moonshotai/kimi-k2.5",
    api_token="env:NOVITA_API_KEY",   # or set NOVITA_API_KEY in environment
    base_url="https://api.novita.ai/openai",
)

Supported models:

Model	Context	Notes
novita/moonshotai/kimi-k2.5	262k	Recommended default
novita/zai-org/glm-5	202k
novita/minimax/minimax-m2.5	204k

For Docker deployments, pass the following environment variables:

NOVITA_API_KEY=your-novita-key
NOVITA_BASE_URL=https://api.novita.ai/openai

The existing get_llm_base_url() utility already picks up NOVITA_BASE_URL via the generic {PROVIDER_NAME}_BASE_URL pattern — no Docker-layer changes were needed.

How it works

crawl4ai routes all LLM calls through litellm. For OpenAI-compatible providers, litellm accepts any model string with an explicit api_base. The PROVIDER_MODELS_PREFIXES entry ensures the API key is resolved automatically from NOVITA_API_KEY whenever a novita/… provider string is used, matching the pattern of existing providers (deepseek, groq, etc.).