Request to enable "Public API Endpoints" permission for my personal organization

AI & Data Science Deep Learning (Training & Inference) TensorRT
, ,
1 
Hello NVIDIA team,

I am unable to call any NIM API models via the NVIDIA API endpoint. All requests return 403 "Authorization failed".

Registered email: 【你的注册邮箱】
API Key prefix: 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

I have followed the official documentation to register and generate my API key, but it seems my personal organization is missing the "Public API Endpoints" permission.

Could you please enable this permission for my account? Thank you for your support.
2

Hi @wangjaifei welcome to the community!

A 403 on the NVIDIA API endpoints usually means one of two things:

  1. Your API key was generated on a personal org that hasn’t completed the free-tier onboarding. Go to build.nvidia.com, sign in, pick any model, and click “Get API Key” from that page. This flow auto-provisions the “Public API Endpoints” permission on your org. Keys generated directly from the NGC Personal Keys page sometimes miss this scope.

  2. Your free credits may be exhausted. Personal orgs get a limited number of free inference credits. Once they are used up, all API calls return 403 until credits are refreshed or you upgrade to a paid plan.

Try generating a fresh key through the build.nvidia.com flow and test with a simple curl:

curl -s -w "\n%{http_code}" \
  https://integrate.api.nvidia.com/v1/chat/completions \
  -H "Authorization: Bearer nvapi-YOUR_NEW_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"meta/llama-3.1-8b-instruct","messages":[{"role":"user","content":"Hello"}],"max_tokens":16}'

If you still get 403, check the response body, it will tell you whether it is an authorization issue or a quota issue.

Important: Please do not share your full API key (or its prefix) publicly. I’d recommend rotating the key you posted above via NGC Personal Keys as a precaution.

3

*** Please note that this reply is generated by LLM automatically ***

TensorRT is a high-performance deep learning inference optimizer and runtime that can be used to optimize and deploy deep learning models on NVIDIA GPUs. Here’s a comprehensive overview of TensorRT:

Key Features:

  1. Optimization: TensorRT can optimize deep learning models by reducing the computational complexity and memory usage, resulting in faster inference times.
  2. Runtime: TensorRT provides a runtime environment that can execute optimized models on NVIDIA GPUs.
  3. Support for Multiple Frameworks: TensorRT supports multiple deep learning frameworks, including TensorFlow, PyTorch, and Caffe.
  4. Dynamic Shape Support: TensorRT supports dynamic shapes, which allows for more flexible and efficient model deployment.
  5. Quantization: TensorRT provides tools for quantizing models, which can further reduce computational complexity and memory usage.
  6. Sparse Tensors: TensorRT supports sparse tensors, which can improve performance and reduce memory usage for models with sparse data.

How TensorRT Works:

  1. Model Optimization: TensorRT optimizes deep learning models by applying various techniques, such as kernel fusion, dead code elimination, and constant folding.
  2. Model Compilation: TensorRT compiles optimized models into a format that can be executed on NVIDIA GPUs.
  3. Runtime Execution: TensorRT executes compiled models on NVIDIA GPUs, providing fast and efficient inference.

TensorRT APIs:

  1. C++ API: TensorRT provides a C++ API that allows developers to optimize and deploy models programmatically.
  2. Python API: TensorRT provides a Python API that allows developers to optimize and deploy models using Python.
  3. TensorRT Engine: The TensorRT engine is a binary format that represents an optimized model. It can be loaded and executed on NVIDIA GPUs.

Comparison with Other Frameworks:

  1. TensorFlow: TensorRT is designed to work with TensorFlow, and provides a TensorFlow-TensorRT compiler that can optimize and deploy TensorFlow models.
  2. PyTorch: TensorRT provides a PyTorch-TensorRT compiler that can optimize and deploy PyTorch models.
  3. OpenVINO: OpenVINO is another deep learning inference optimizer and runtime that competes with TensorRT. While both frameworks provide similar functionality, TensorRT is specifically designed to work with NVIDIA GPUs.

Best Practices:

  1. Use the Latest Version: Always use the latest version of TensorRT to take advantage of new features and performance improvements.
  2. Optimize Models: Optimize models using TensorRT’s optimization tools to reduce computational complexity and memory usage.
  3. Use Dynamic Shapes: Use dynamic shapes to allow for more flexible and efficient model deployment.
  4. Quantize Models: Quantize models to further reduce computational complexity and memory usage.
  5. Use Sparse Tensors: Use sparse tensors to improve performance and reduce memory usage for models with sparse data.

Common Use Cases:

  1. Deep Learning Inference: TensorRT is commonly used for deep learning inference, where optimized models are deployed on NVIDIA GPUs for fast and efficient execution.
  2. Computer Vision: TensorRT is widely used in computer vision applications, such as image classification, object detection, and segmentation.
  3. Natural Language Processing: TensorRT is used in natural language processing applications, such as language translation and text classification.
  4. Autonomous Vehicles: TensorRT is used in autonomous vehicle applications, such as object detection and tracking.

*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***