NVIDIA NIM is a series of microservices that it offers preconfigured and optimized avant -garde AI models For the NVIDIA RTX platform, including GEFORCE RTX 50 consumer graphics cards and Blackwell RTX Pro 6000 professional cards. Microservices are now available and stand out for their simplicity for download and execution. They cover the main development modalities for PC and are compatible with the main applications and tools of the green giant ecosystem.
Nvidia continues to produce products and services at a good pace. If last week we saw a large number of novelties in its GTC 2025 conference, the company has just announced the launch of other solutions that focus on what is becoming its main business, applications that dominate current technology such as AI and deep learning. One of the novelties has been the Project G-Assist An intelligent personal assistant enhanced by an AI model based on the flame that our MC partners have presented us. Another of the great novelties has been the launch of NIM.
What is Nvidia Nim
At a crucial time for computer science, where AI models and a global developer community drive an explosion of tools and workflows based on AI, NIM microservices want to help key innovations to personal computers. Although the rhythm of innovation with AI is incredible, it can still be difficult for the PC developer community to start in these technologies.
And it is that carrying the models of the research to the PC requires the selection of variants of the model, the adaptation to manage all the input and output data, and the quantification to optimize the use of resources. In addition, the models must become to work with software backend of optimized inference and connect to new application programming interfaces. This requires considerable effort, which can slow the adoption of artificial intelligence.
NVIDIA NIM microservices help solve this problem by providing Models of preconfigured, optimized and easy to download that connect to the standard APIs of the sector. They are optimized for professional PCs and workstations with the company’s RTX hardware and include the best AI models in the community, as well as models developed by NVIDIA.
NIM microservices are compatible with various AI applications, such as large language models (LLM), vision language models, image generation, voice processing, recovery-based search-generation (RAG), PDF extraction and artificial vision.
Currently, there are Ten nvidia NIM microservices available for RTXcompatible with various applications, such as language and images generation, artificial vision, voice, and specifically the following:
- Language and reasoning: Deepseek-R1-Distill-Llama-8b, Mistral-Nemo-12B-Instruct, Call3.1-8b-Instruct
- Image generation: Flux.dev
- Audio: Riva Parakeet-ctc-0.6B-asr, Maxine Studio Voice
- RAG: Llama-3.2-NV-EmbedQA-1B-v2
- Computer vision: NV-clip, paddleoc, yolo-X-v1
NIM microservices are also available through main tools and frameworks of the ecosystem of AI. For example, Anythingllm and Chatrtx are now compatible with NIM, which facilitates chat with LLM and AI agents through a simple and intuitive interface. With these tools, users can create personalized AI attendees and integrate their own documents and data, which helps automate tasks and improve productivity. Microsoft vs Code Ai Toolkit, Crewai and Langchain are also compatible, providing advanced abilities to integrate microservices into the application code, which helps guarantee integration and optimization.
NVIDIA AI Blueprints
The company has also presented another solution that offers the developers of AI an advantage in the Creation of generative work flows with microservices Nvidia Nim. Blueprints are extendable reference samples and ready to use that include everything necessary (source code, documentation data and a demonstration application) to create and customize advanced workflows that are executed locally.
Developers can modify and expand the AI planes to adjust their behavior, use different models or implement completely new functionalities.
More information | Nvidia Developer
