# Together AI MCP for AI Agents AI Agent Connect > Together AI MCP lets you run 100+ open-source models like Llama 3.3 and Flux directly through your AI client. It gives you the tools to handle text generation, image creation, audio transcription, and high-scale batch processing without managing any underlying hardware. This is the fastest way to get production-ready inference for your own applications. ## Overview - **Category:** ai-frontier - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_3DCNpnHYF8hhWcXn2SCKuEfpGwWZe18MoQz7sb18/ai-agent-connect - **Tags:** llm, generative-ai, llama-3, image-generation, fine-tuning ## Description This Connector connects your AI agent to the Together AI inference cloud. Instead of worrying about hosting heavy models, you can just ask your agent to run a prompt through the world's fastest inference cloud for open-source models. It handles the heavy lifting for text completions and batch processing, so you can focus on the final output. If you need to build a custom search system, it can generate vector numbers or reorder your results to improve accuracy. You can even manage your own fine-tuning jobs and dedicated endpoints from a single interface. It's a way to get production-grade performance from open-source models without the headache of setting up GPU clusters. By connecting this to the Vinkius catalog, you get a unified way to hit these models whenever your agent needs to do something complex. You get the power of the latest weights without the infrastructure debt. It makes it easy to scale from a single prompt to a massive batch of requests in seconds. ## Tools ### cancel_batch Stop a batch job that's running. This helps if you need to kill a task early. ### create_chat_completion Get a response from a chat model. Use this for standard conversational AI tasks. ### create_batch Start an asynchronous batch job. This is the way to handle large volumes of data at once. ### list_fine_tunes See all your current fine-tuning jobs. This gives you a bird's eye view of your training. ### create_audio_speech Turn text into spoken audio. It's great for making your AI agent talk. ### create_audio_transcription Turn audio files into text. Use this to get transcripts with speaker IDs. ### create_endpoint Set up a dedicated endpoint. Use this when you need consistent, predictable performance. ### create_fine_tune Start a new fine-tuning job. This lets you train a model on your specific data. ### delete_endpoint Remove a dedicated endpoint. Use this to clean up your resources when you're done. ### delete_file Remove an uploaded file. This keeps your storage clean after a job is finished. ### delete_fine_tune Delete a finished fine-tuning job. This helps manage your active training projects. ### create_embeddings Turn text into vector numbers. This is how you build a search system for your documents. ### get_batch Check the status of a batch job. Use this to see if your large task is finished. ### get_endpoint See the details of a dedicated endpoint. This helps you monitor your custom hardware setup. ### get_file See the metadata for a specific file. Use this to check if your upload was successful. ### get_fine_tune Check the progress of a fine-tuning job. This lets you see how your training is going. ### create_image_generation Create an image from a text prompt. Use this for generating visual content on the fly. ### list_endpoints See all your dedicated endpoints. This helps you keep track of your active hardware. ### list_files See all the files you've uploaded. Use this to manage your training data. ### list_fine_tune_checkpoints See the progress points for a fine-tune job. This is useful for monitoring training. ### list_models See all the models available on Together AI. Use this to find the best model for your task. ### create_rerank Reorder search results by relevance. This makes your search systems much more accurate. ### create_text_completion Get text based on a prompt. Use this for simple completions without a full chat history. ### update_endpoint Start, stop, or scale a dedicated endpoint. This gives you control over your performance. ### upload_file Send a file to the cloud. Use this to provide data for fine-tuning or batch jobs. ### create_video_generation Make a video from a prompt or image. This is the way to generate motion content. ### list_batches See all your active batch jobs. This helps you manage your asynchronous workloads. ## Prompt Examples **Prompt:** ``` Can you generate a summary of this long article using Llama 3.3? ``` **Response:** ``` ### Article Summary Based on the text provided, here are the key takeaways: * **Main Theme:** The impact of decentralized finance on local economies. * **Key Points:** * Increased accessibility for rural users. * Challenges with regulatory compliance. * The role of stablecoins in inflation hedging. **Conclusion:** The article suggests that while hurdles remain, the shift toward decentralized systems is accelerating faster than predicted. ``` **Prompt:** ``` Create a futuristic city image using Flux. ``` **Response:** ``` Generating image... I've successfully created the image of a futuristic city. **Image Details:** * **Style:** Cyberpunk, high-detail * **Elements:** Neon lights, flying vehicles, lush vertical gardens * **Lighting:** Twilight with deep purples and oranges You can view the generated image in the gallery or download it via the link provided. ``` **Prompt:** ``` List all the models I can use for text generation. ``` **Response:** ``` You have access to a wide range of high-performance models for text generation. Here are some of the most popular options: | Model | Best For | | :--- | :--- | | **Llama 3.3** | General purpose chat and complex reasoning | | **Mixtral** | Efficient, high-speed text completion | | **Qwen** | Multilingual tasks and coding | | **Command R** | RAG and long-context tasks | Would you like me to run a test prompt on any of these? ``` ## Capabilities ### Generate text responses Use Llama 3.3 or Mixtral to get high-quality chat completions. ### Create images from text Use Flux or Stable Diffusion to turn prompts into visual assets. ### Transcribe audio files Turn spoken words into text with speaker identification. ### Generate speech Convert your written text into natural-sounding audio. ### Build search systems Create vector embeddings or rerank documents for better search. ### Run large-scale jobs Handle asynchronous tasks using the Batch API. ### Train custom models Manage fine-tuning jobs and checkpoints for your own data. ## Use Cases ### Building a custom chatbot A dev asks their agent to use create_chat_completion with Llama 3.3 to answer customer questions accurately. ### Generating marketing assets A social media manager asks the agent to use create_image_generation to make 50 unique product photos. ### Processing large datasets A researcher uses create_batch to run a text completion on 10,000 rows of data overnight. ### Building a private search A librarian uses create_embeddings to index a library and create_rerank to help users find specific books. ## Benefits - Access 100+ models including Llama 3.3 and Mixtral without managing any hardware. - Generate high-quality images using Flux or Stable Diffusion through a single API. - Build better RAG systems by using create_embeddings and create_rerank for precise search. - Handle massive workloads efficiently by using create_batch to process data in the background. - Train your own custom models with create_fine_tune and manage the process with list_fine_tune_checkpoints. - Get predictable performance for your app by setting up dedicated endpoints with create_endpoint. ## How It Works The bottom line is you get instant access to a massive library of open-source models without managing a single server. 1. Subscribe to the Together AI MCP on Vinkius. 2. Add your Together AI API key to your AI client. 3. Start asking your agent to run models or manage fine-tuning jobs. ## Frequently Asked Questions **Can I use Together AI MCP to run Llama 3.3?** Yes. You can use this Connector to run Llama 3.3 directly through your AI agent for high-quality chat and text completion tasks. **How does Together AI MCP handle image generation?** It connects you to models like Flux and Stable Diffusion, allowing your agent to turn text prompts into high-quality images instantly. **Can I use Together AI MCP for batch processing?** Yes. You can use the batch tools to handle large-scale, asynchronous workloads like processing thousands of text completions at once. **Does Together AI MCP support fine-tuning?** Yes. This Connector allows you to manage your own fine-tuning jobs, create jobs, and monitor checkpoints for your custom models. **Can I use Together AI MCP to build a RAG system?** Absolutely. You can use it to generate vector embeddings and reorder your search results to build a high-performance retrieval system. **How do I get predictable performance with Together AI MCP?** You can set up dedicated endpoints through this Connector to ensure consistent performance for your production applications. **How do I generate a chat response using a specific model like Llama 3.3?** Use the `create_chat_completion` tool. Specify the model name (e.g., 'meta-llama/Llama-3.3-70B-Instruct-Turbo') and provide an array of messages. The agent will return the generated response from the model. **Can I create images from text prompts with this server?** Yes! Use the `create_image_generation` tool. You can specify the model, the prompt description, and optional parameters like width, height, and steps to get high-quality visual outputs. **How can I check the status of my asynchronous batch jobs?** You can use `list_batches` to see all your current batch jobs or `get_batch` with a specific Job ID to retrieve detailed status and results for a particular task.