How to enable Nvidia's Developer AI API on Chatwise

,
1

Jina DeepSearch | ChatWise Docs

  1. Get an NVIDIA Developer (API Catalog / build.nvidia.com) API key
  • Generate an API key from NVIDIA’s API Catalog (often labeled NVIDIA_API_KEY / NIM_API_KEY). (LangChain Docs)
  1. Add NVIDIA as an OpenAI-Compatible provider in ChatWise
  • ChatWise → Settings → Providers → + → OpenAI Compatible / Custom Provider. (docs.chatwise.app)
  • Set API Base URL to: https://integrate.api.nvidia.com/v1 (docs.api.nvidia.com)
  • Paste your NVIDIA API key into API Key (ChatWise will send it as Authorization: Bearer <key> for OpenAI-compatible providers; NVIDIA endpoints use Bearer auth). (docs.api.nvidia.com)
  1. Add (or fetch) model IDs
  • In the same provider config, use Fetch models or add models manually. (docs.chatwise.app)
  • Example model IDs available on NVIDIA’s serverless OpenAI-compatible endpoint:
    • deepseek-ai/deepseek-r1
    • meta/llama3-70b-instruct
    • microsoft/phi-3-mini-128k-instruct
    • mistralai/mistral-7b-instruct-v0.3 (docs.api.nvidia.com)
  1. Use it
  • Select the NVIDIA provider/model in ChatWise and start a normal chat (ChatWise calls /v1/chat/completions under that base URL). (docs.api.nvidia.com)

Common failures

  • 401 Unauthorized: key invalid/expired or copied with whitespace → regenerate key. (NVIDIA Developer Forums)
  • 404 / not found: base URL wrong (must be .../v1, not .../v1/chat/completions in the base field). (docs.api.nvidia.com)
  • Slow/queued responses: serverless trial can have high-load wait times. (NVIDIA Developer Forums)