# NVIDIA AI MCP for AI Agents AI Agent Connect > NVIDIA AI lets your AI agent access the full NVIDIA API Catalog. It gives your agent the ability to run high-performance models like Llama 3.1 and Mistral, generate embeddings, write code, and perform complex reasoning using GPU-accelerated infrastructure without you having to manage a single server. ## Overview - **Category:** industry-titans - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_F8wZEFp9XAw3aowvQGCJgs0iHR9Eswli1t6PkJLP/ai-agent-connect - **Tags:** llm, gpu-acceleration, embeddings, model-inference, natural-language-processing, code-generation ## Description Instead of worrying about how to host heavy models or managing your own GPU clusters, this Connector lets you plug directly into production-grade infrastructure. You can have your agent handle everything from basic conversations to heavy-duty tasks like converting natural language into database queries or generating production-ready code. It is about getting the right model for the right job. Whether you need a high-parameter reasoning model for complex logic or a dedicated model for your search index, you can do it here. If you are looking for a way to move fast on AI features without the headache of infrastructure, this is how you do it. You can find this in the Vinkius catalog to get your agent connected to these models in minutes. ## Tools ### ask_question Ask a high-parameter reasoning model a complex question. You can provide extra context to get a more nuanced answer. ### chat_completion Start a conversation with models like Llama or Mistral. You just need to specify the model name and the user messages. ### generate_code Turn a description of a coding task into actual source code. It works for multiple programming languages. ### get_embeddings Turn a block of text into a vector embedding. This helps with building search and clustering systems. ### list_models See every model currently available in the NVIDIA API Catalog. It helps you pick the right tool for your specific task. ### text_to_sql Give the agent a natural language question and get a SQL query back. It makes it easier to talk to your database. ### analyze_sentiment Check the emotional tone of a piece of text. This is great for monitoring feedback or reviews. ### summarize_text Turn a long document into a short summary. It helps you get the main points without reading the whole thing. ### translate_text Convert text from one language to another. It supports dozens of different languages for global reach. ## Prompt Examples **Prompt:** ``` Can you write a Python script to scrape a website and save it to a CSV? ``` **Response:** ``` **Generated Code:** ```python import requests from bs4 import BeautifulSoup import csv def scrape_site(url): response = requests.get(url) soup = BeautifulSoup(response.text, 'html.parser') # Example logic for extracting data data = soup.find_all('p') with open('data.csv', 'w', newline='') as f: writer = csv.writer(f) for item in data: writer.writerow([item.text]) scrape_site('https://example.com') ``` ``` **Prompt:** ``` What is the sentiment of this review: 'The delivery was late, but the product quality is absolutely incredible.' ``` **Response:** ``` The sentiment of the review is **Mixed/Positive**. * **Negative:** Delivery delay. * **Positive:** High product quality. Overall, the customer seems satisfied with the purchase but frustrated with the logistics. ``` **Prompt:** ``` Translate 'Where is the nearest train station?' to Japanese. ``` **Response:** ``` 最寄りの駅はどこですか？ *(Moyori no eki wa doko desu ka?)* ``` ## Capabilities ### Run chat completions Connect your agent to models like Llama and Mistral for conversational tasks. ### Create vector embeddings Turn text into vectors to power your search and clustering systems. ### Generate production code Turn natural language prompts into functional code across multiple languages. ### Convert text to SQL Let your agent translate plain English questions into executable database queries. ### Summarize long documents Condense large amounts of text into concise summaries automatically. ### Translate languages Convert text between dozens of different languages with neural translation. ### Analyze sentiment Identify the emotional tone of text for feedback and review analysis. ### Perform complex reasoning Use a 405B parameter model to solve difficult logic and reasoning problems. ## Use Cases ### Building a RAG system A data scientist needs to process 10,000 documents for a search index. They use get_embeddings to create a high-quality vector store. ### Automating database reports A business analyst asks the agent to find last month's sales. The agent uses text_to_sql to query the database and return the data. ### Multi-language customer support A support lead wants to handle global inquiries. They use translate_text to convert customer messages into English for the team. ### Rapid prototyping of AI features A developer wants to test a chatbot quickly. They use chat_completion with Mistral to verify the logic before choosing a final model. ## Benefits - Skip the GPU headache by using NVIDIA's infrastructure for all your model hosting needs. - Get high-quality code generation by using generate_code for specific programming tasks. - Build faster search features by using get_embeddings to create high-quality vector data. - Query your databases in plain English by using the text_to_sql tool to automate data retrieval. - Handle complex logic with the 405B reasoning model for tasks that smaller models struggle with. - Run diverse models like Mistral and Llama 3.1 through a single connection point. ## How It Works The bottom line is you get production-grade GPU models ready for your agent instantly. 1. Subscribe to the NVIDIA AI MCP via the Vinkius marketplace. 2. Add your NVIDIA API Key from build.nvidia.com to your configuration. 3. Start asking your agent to run models from Claude, Cursor, or other clients. ## Frequently Asked Questions **Does NVIDIA AI support Llama 3.1 models?** Yes, you can access Llama 3.1 and other high-performance models directly through the NVIDIA API Catalog using this Connector. **Can I use NVIDIA AI to build a RAG system?** Yes, you can use the embedding tools to convert your data into vectors, which is a core requirement for building RAG systems. **How do I get my NVIDIA API Key?** You can generate your API key at the official NVIDIA build website and add it to your Connector configuration. **Can NVIDIA AI write SQL queries for me?** Yes, the text-to-SQL tool allows your agent to take a natural language question and turn it into a valid SQL query for your database. **What models are available through NVIDIA AI?** You can see the full list of available models, including Llama, Mistral, and Nemotron, by using the model listing tool. **Is this Connector good for translation tasks?** Yes, it supports neural translation between dozens of different languages, making it great for global content needs. **Which AI models are available?** The NVIDIA API Catalog offers Llama 3.1 (8B, 70B, 405B), Mistral, CodeLlama, Gemma, Nemotron, and many more. Use the `list_models` tool to see all available models. **How do I get an NVIDIA API Key?** Sign up at [**build.nvidia.com**](https://build.nvidia.com), go to your account settings, and generate an API key. The Developer Program includes free inference credits. **Can I generate code in specific languages?** Yes! The `generate_code` tool lets you specify the programming language (Python, JavaScript, TypeScript, Java, etc.) for better results. **Are there usage limits on the free tier?** Yes, the NVIDIA Developer Program provides free inference credits. Once exhausted, you can upgrade to a paid plan for higher throughput. Check your usage dashboard at build.nvidia.com.