Popularity metric used for sorting: catalog breadth (how many models/endpoints NVIDIA exposes under that API family in the NVIDIA NIM API Reference).
| Popularity (proxy) | NVIDIA AI API family | Base domain | Canonical endpoint pattern(s) | What it covers |
|---|---|---|---|---|
| 1 | LLM APIs (OpenAI-compatible) | integrate.api.nvidia.com |
POST /v1/chat/completions ([docs.api.nvidia.com][1])POST /v1/completions ([docs.api.nvidia.com][2]) |
Text chat/instruction/code generation via OpenAI-style APIs (model chosen via model field). ([docs.api.nvidia.com][3]) |
| 2 | Visual Models APIs | ai.api.nvidia.com |
POST /v1/genai/{publisher}/{model} (e.g., .../black-forest-labs/flux.1-dev) ([docs.api.nvidia.com][4])POST /v1/cv/nvidia/visual-changenet ([docs.api.nvidia.com][5]) |
Image generation and computer-vision model endpoints (path encodes the specific model/service). ([docs.api.nvidia.com][4]) |
| 3 | Healthcare APIs | health.api.nvidia.com |
POST /v1/biology/deepmind/alphafold2 ([docs.api.nvidia.com][6])POST /v1/medicalimaging/nvidia/maisi ([docs.api.nvidia.com][7]) |
Biology + medical imaging model endpoints (domain/path encodes the specific service). ([docs.api.nvidia.com][6]) |
| 4 | Retrieval APIs | integrate.api.nvidia.com + ai.api.nvidia.com |
POST /v1/embeddings ([Elastic][8])POST /v1/retrieval/nvidia/embeddings ([docs.api.nvidia.com][9])POST /v1/retrieval/nvidia/reranking ([docs.api.nvidia.com][10]) |
Embeddings + reranking endpoints (OpenAI-compatible embeddings on integrate, plus retrieval endpoints on ai.api). ([Elastic][8]) |
| 5 | Multimodal APIs | integrate.api.nvidia.com |
POST /v1/chat/completions (multimodal-capable models) ([docs.api.nvidia.com][1]) |
Multimodal inference uses the same OpenAI-compatible chat endpoint; large image workflows reference NVCF asset flow in the docs. ([docs.api.nvidia.com][11]) |
| 6 | Climate Simulation APIs | climate.api.nvidia.com |
POST /v1/nvidia/fourcastnet ([docs.api.nvidia.com][12])POST /v1/nvidia/corrdiff ([docs.api.nvidia.com][13]) |
Weather/climate inference endpoints (Earth-2). ([docs.api.nvidia.com][12]) |
| 7 | Route Optimization APIs | optimize.api.nvidia.com |
POST /v1/nvidia/cuopt ([docs.api.nvidia.com][14])GET /v1/nvidia/cuopt (status polling) ([docs.api.nvidia.com][15]) |
cuOpt solver submission + status polling. ([docs.api.nvidia.com][14]) |
| — | Execution/async support (NVCF) | api.nvcf.nvidia.com |
GET /v2/nvcf/pexec/status/{req_id} ([docs.api.nvidia.com][13]) |
Status polling for async executions referenced by some NVIDIA endpoints/examples. ([docs.api.nvidia.com][13]) |
Chatwise/OpenAI-client compatible base URL (for NVIDIA LLM/multimodal via OpenAI-style endpoints): https://integrate.api.nvidia.com/v1 ([docs.api.nvidia.com][1])