feat: add Novita AI as LLM provider

[Alex-wuhu]

Summary

Add Novita AI as a new LLM provider. Novita offers an OpenAI-compatible API with competitive pricing and a wide selection of open-source models.

Changes

• New Novita provider following existing provider patterns
• API key configuration via NOVITA_API_KEY environment variable
• Support for chat, completion, and embedding models

Configuration

NOVITA_API_KEY=your_key_here

Endpoint: https://api.novita.ai/openai

[vercel]

@Alex-wuhu is attempting to deploy a commit to the Elie Team on Vercel.

A member of the Team first needs to authorize it.

[SebastienMelki]

@Alex-wuhu — implementation follows the existing provider pattern cleanly. A few questions:

1. Branch has merge conflicts — please rebase on main
2. Are there Novita free-tier rate limits we should document in .env.example?
3. Is https://api.novita.ai/openai the stable production endpoint or might it change?
4. Does Novita support streaming responses? Matters for the AI Brief path.

[SebastienMelki]

Note: merge conflicts with main. Fork PR — can't rebase from maintainer side. @Alex-wuhu — please rebase on current main when addressing the review questions.

[Alex-wuhu]

Hi @SebastienMelki, thanks for the review! Here are the answers:

1. Rebase on main — Done. Conflicts resolved and pushed.

2. Novita free-tier rate limits for .env.example?
Novita AI is purely pay-per-use with no free tier. Neither the pricing page nor the API docs mention request rate limits. So there's nothing to document like Groq's "14,400 req/day" or OpenRouter's "50 req/day" — the current .env.example comment is sufficient.

3. Is https://api.novita.ai/openai the stable production endpoint?
Yes. The official docs explicitly list https://api.novita.ai/openai as the production base URL. The full chat completions path used in the PR (https://api.novita.ai/openai/v1/chat/completions) is correct and stable.

4. Does Novita support streaming?
Yes — the docs state it "supports both streaming and non-streaming modes" via the standard stream: true parameter (SSE). Since callLlmReasoningStream handles streaming generically for all OpenAI-compatible providers, Novita works on the AI Brief path without any additional code.

---

Also — would you be interested in doing a co-post on X with Novita AI about this integration? Let us know!