NVIDIA NIM

Overview

NVIDIA NIM (NVIDIA Inference Microservices) provides GPU-accelerated serving for a broad catalog of models, optimized for high throughput and low latency.

Official Website: https://build.nvidia.com Documentation: https://docs.nvidia.com/nim

Key features

• GPU acceleration — Optimized, high-performance inference
• Chat & vision — Text generation and image understanding on supported models
• Function calling — Tool use support
• Embeddings — Text embeddings for search and retrieval
• Image generation — Create images from text
• 3D generation — Text-to-3D and image-to-3D
• Reranking — Rerank candidate documents for retrieval pipelines

Using NVIDIA NIM with Yunxin

Address NVIDIA models with the nvidia/<model> prefix in the model field. You use the same Yunxin base URL (https://api.yuhuanstudio.com/v1) and your sk- API key — there is no separate NVIDIA configuration on your side.

Capabilities vary by model. Query GET /v1/models to discover available models and what each one supports.

Usage example

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["YUNXIN_API_KEY"],
    base_url="https://api.yuhuanstudio.com/v1"
)

response = client.chat.completions.create(
    model="nvidia/model-id",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ]
)

print(response.choices[0].message.content)

Available models

Query the Models API for the current list and per-model capabilities:

curl "https://api.yuhuanstudio.com/v1/models?provider=nvidia" \
  -H "Authorization: Bearer $YUNXIN_API_KEY"

Official resources

• NVIDIA Build
• NIM Documentation
• Model Catalog