The Together AI MCP for open-weight models inside your automations
Connect one Together AI API key and chat completions on Llama, Qwen, and DeepSeek, FLUX image generation, embeddings, document rerank, Whisper audio, batch jobs, and fine-tuning become steps on a visual canvas that AI agents can drive. 22 typed nodes over Together's OpenAI-compatible API.
What the Together AI MCP connector does
The Together AI connector for Mattering Automations covers Together's documented inference surface as 22 typed nodes: chat and raw text completions on open-weight models, embeddings, document rerank, image generation with FLUX, Whisper transcription and translation from an audio URL, asynchronous batch jobs, file listing, metadata, content download and delete, and a six-node fine-tuning lifecycle. Nine are read-only queries and 13 are actions.
Every node speaks MCP, the open protocol for connecting AI assistants to tools. Claude, ChatGPT, and any MCP-capable assistant can call them, and so can Matter, the assistant built into the platform. That is the point of the Together AI MCP: an assistant plans the workflow, hands the generation, the image, or the bulk batch to a Together-hosted open model, and both sit on one canvas you can see and edit.
Two Together-specific details shape the node list. First, every model field is free text with a documented default rather than a fixed dropdown, because Together's serverless catalog changes constantly; Together AI: List Models returns everything your key can call, with each model's type, context length, and per-token pricing, and since Together documents no single-model lookup, that list is the lookup surface. Second, image generation is pinned to URL responses on purpose: the node returns a hosted image URL instead of inline base64, so a workflow that generates images keeps its execution records small and pipes the URL to the next step.
One connection, three surfaces
- Visual automations. Drag Together AI nodes onto the canvas and wire them to triggers, conditions, and other apps.
- AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
- The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.
Every Together AI node, ready to drop on the canvas
22 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.
Read and query nodes 9
Together AI: Download File Content
Download a file's contents as text. Batch output and error files are JSONL, so the node also returns each line parsed into a lines array you can loop over.
Together AI: Get Batch
Get one batch job: status, progress percentage, and the output and error file IDs to download once it completes.
Together AI: Get File
Get one file's metadata: size in bytes, filename, purpose, and processing status.
Together AI: Get Fine-Tuning Job
Get one fine-tuning job: status, epochs completed, token count, and the model_output_name you pass as model on a chat completion once training finishes.
Together AI: List Batches
List the batch jobs on the account with status (VALIDATING, IN_PROGRESS, COMPLETED, FAILED, EXPIRED, CANCELLED) and progress.
Together AI: List Files
List the files uploaded to the account, with size, filename, purpose, and processing status.
Together AI: List Fine-Tuning Events
List the event log for a fine-tuning job: pending, training start, checkpoint saves, completion, or errors. Useful for polling training progress from a flow.
Together AI: List Fine-Tuning Jobs
List the fine-tuning jobs on the account with status, base model, and output model name.
Together AI: List Models
List every model available to the API key, with type (chat, language, image, embedding, rerank, moderation), context length, and per-token pricing. Works with no parameters, so it doubles as the credential health check. Together documents no single-model GET, so this list is the lookup surface.
Action nodes 13
Together AI: Cancel Batch
Cancel a batch job that is still validating or in progress.
Together AI: Cancel Fine-Tuning Job
Cancel a running fine-tuning job. Billing stops for work not yet performed.
Together AI: Create Batch
Kick off an asynchronous batch job over a previously uploaded JSONL file. Batch runs are discounted and write results to an output file when they finish. The response wraps the job under data.job.
Together AI: Create Chat Completion
Run a chat completion on a Together-hosted open-weight model (Llama, Qwen, DeepSeek, Mixtral, and more). Send a messages array, get the assistant reply at choices[0].message.content plus token usage. Supports tool calling and JSON-mode structured output. Non-streaming.
Together AI: Create Completion
Run a raw text completion: send a prompt string, get the continuation at choices[0].text. Use this for base language models or when you want to control the prompt template yourself; use Create Chat Completion for instruct models. Non-streaming.
Together AI: Create Embedding
Turn text into embedding vectors for search, clustering, or RAG. Send one string or a JSON array of strings; each input comes back as a float vector at data[n].embedding.
Together AI: Create Fine-Tuning Job
Start a fine-tuning job (LoRA by default, full fine-tuning optional) on an uploaded JSONL training file. This launches paid GPU training billed by token count, so check pricing before wiring it into an automation. The training file must be uploaded out-of-band (CLI or SDK) first.
Together AI: Delete File
Permanently delete an uploaded file from the Together account.
Together AI: Delete Fine-Tuning Job
Delete a fine-tuning job record from the account. This cannot be undone.
Together AI: Generate Image (URL)
Generate an image from a text prompt with FLUX or another Together-hosted image model. The result comes back as a hosted image URL at data[0].url, never inline base64, so pipe the URL onward or download it promptly if you need to keep the image.
Together AI: Rerank Documents
Rank a list of documents by relevance to a query. Returns each document's index and relevance_score, best first. Note: Together currently serves rerank models on dedicated endpoints rather than serverless, so check List Models for what your account can call.
Together AI: Transcribe Audio (URL)
Transcribe an audio file into text with Whisper on Together. Takes a publicly reachable audio URL (Together fetches it, up to 1 GB); this node does not upload local files. Supports speaker diarization. Formats: wav, mp3, m4a, webm, flac, ogg, opus, aac.
Together AI: Translate Audio to English (URL)
Translate non-English audio into English text with Whisper on Together. Takes a publicly reachable audio URL (up to 1 GB); this node does not upload local files.
Connecting Together AI takes one credential
Together authenticates with a single API key created at api.together.ai under Settings and API Keys. There is no second field, no organization header, and no OAuth flow to complete, and new Together accounts come with trial credit, so a fresh key can run real calls before any payment is set up.
At api.together.ai, open Settings, then API Keys, create a key, and copy it.
In Mattering Automations, open Connectors, choose Together AI, and create a credential holding the key.
Click Connect. All 22 Together AI nodes appear in the canvas palette.
Drop Together AI: List Models on a canvas and run it as a test step. It takes no parameters, costs no tokens, and returns every model your key can call, so it doubles as the credential health check.
Common uses for the Together AI MCP
Generation steps on open-weight models
Drafting, rewriting, classifying, and structured extraction as one node between your data source and its destination, running on Llama, Qwen, DeepSeek, or any other model in Together's catalog. The chat completion node supports tool calling, JSON-mode structured output, and Together-specific controls such as a Llama-Guard safety pass and a truncate option when a prompt exceeds the context window.
Images generated mid-workflow
Together AI: Generate Image (URL) turns a text prompt into a FLUX image and hands back a hosted URL, so the very next node can post it, attach it, or download it into your own storage. Because the response is a URL rather than inline image data, an automation that produces dozens of images stays light.
Vectors and rerank for search
Together AI: Create Embedding turns one string or a whole array of them into float vectors for search, clustering, or RAG, and Together AI: Rerank Documents reorders a candidate list by relevance to a query, best first. Rerank models currently run on dedicated endpoints rather than serverless, so check List Models for what your account can call.
Audio to text without a separate transcription vendor
Point Together AI: Transcribe Audio (URL) at a recording's URL and get the text back with Whisper, with speaker diarization available, or use the translation node to turn non-English audio into English text in the same step. Together fetches the audio itself from URLs up to 1 GB.
Bulk work at the batch discount
Start a batch over an uploaded JSONL file, poll it for status, then download the output file. Batch runs are discounted, and the download node returns the raw text plus each JSONL line already parsed into a lines array, so a For Each node can loop the results straight away.
A fine-tuning loop you can watch from a flow
Kick off a LoRA or full fine-tune on an uploaded training file, poll Together AI: List Fine-Tuning Events for checkpoint saves and completion, then pass the returned model name as the model on a chat completion. Creating a job launches paid GPU training billed by trained tokens, so check Together's pricing before wiring it into anything scheduled.
Example: A product row becomes launch copy and a launch image
Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.
Trigger: a new row lands in a Databases table with a product name and a short feature list.
Together AI: Create Chat Completion turns the row into launch copy, with the row's fields dropped straight into one entry of the messages array as template references.
Together AI: Generate Image (URL) renders a FLUX image from a prompt built on the same fields and returns a hosted image URL.
The workflow downloads the image into your own storage, writes the copy and the stored image back onto the row, and posts both to the team, so the launch asset exists the moment the row does.
Together AI MCP questions, answered
Can an AI agent run open-weight models through the Together AI MCP?
Yes. Matter and any MCP-connected assistant, including Claude and ChatGPT, can run chat and text completions, generate images, create embeddings, rerank documents, transcribe audio, submit and poll batch jobs, and manage files and fine-tuning jobs on the account. They use the same 22 nodes the automation canvas uses.
Does this work with Claude, ChatGPT, or other AI assistants?
Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant can connect to your workspace and reach Together AI through them. Matter, the assistant built into the platform, uses the same nodes natively. Nothing here is tied to one assistant: the point of the connector is to give whichever assistant you already work with access to Together's open-weight models.
Which models can these nodes call?
Whatever your Together API key has access to. Every model field is free text with a documented default, such as meta-llama/Llama-3.3-70B-Instruct-Turbo on chat and black-forest-labs/FLUX.1-schnell on images, rather than a fixed dropdown, because Together's serverless catalog changes on a rolling basis. Run Together AI: List Models for the current set, with each model's type, context length, and per-token pricing. Together documents no single-model lookup endpoint, so that list is also how you check one model's details.
Why does image generation return a URL instead of the image itself?
By design. The connector pins Together's image response format to URL, so the node returns a hosted link at data[0].url and never inline base64. That keeps every execution record small, which matters once a workflow generates images in volume, and a URL pipes cleanly into whatever comes next. Together does not document how long the hosted URL stays live, so if the image matters beyond the run, add a step that downloads it into your own storage.
Can I fine-tune a model from an automation?
Yes, and unlike some providers Together does not gate fine-tuning behind an enterprise tier: any API key can use the six lifecycle nodes. Be deliberate with it, because Together AI: Create Fine-Tuning Job launches real paid GPU training billed by trained tokens. The training file has to be uploaded out of band with Together's CLI or SDK first; from the file id onward the whole lifecycle runs in the workflow, through to passing the finished model's name as the model on a chat completion.
Why does the rerank node say the model is not available?
Together currently serves rerank models on dedicated endpoints rather than serverless, so Together AI: Rerank Documents works on accounts that have a dedicated endpoint running a rerank model and returns a model availability error on accounts that do not. Together AI: List Models takes a dedicated filter, which shows exactly what your account can call. The node ships anyway because the API is documented and works where the endpoint exists.
What is not covered by the 22 nodes?
Three deliberate gaps. File upload needs a real binary multipart payload and the canvas config panel has no file input, so uploads happen out of band; the list, metadata, download, and delete half of the files API is shipped, which means batch and fine-tune workflows still run end to end from a file id onward. Text to speech returns raw audio bytes rather than JSON, and this connector has nowhere to put those bytes yet. And streaming is out of scope: nodes return one complete response, and the connector strips stream from outbound bodies so a payload pasted over from OpenAI cannot silently break a node.
What credential does the connector use, and who bills the usage?
One Together AI API key, sent as a bearer token and stored encrypted in Google Secret Manager. Together bills the tokens, images, audio minutes, batch jobs, and fine-tuning to that key on your own Together account, under the pricing and limits you hold there. That is separate from Mattering credits, which are pooled at the organization level and cover running the automation itself.
Put Together AI to work inside your workspace
Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Together AI is a product of its respective owner; Mattering is an independent platform.