Together AI MCP server

What the model can do

Together AI platform operations: fine-tuning jobs with status, progress and events, batch inference jobs, dedicated endpoints with their deployments and replicas, GPU clusters, uploaded files (by id and purpose) and evaluation jobs; when you ask, can start a batch job (billed per token), cancel a batch or fine-tune, scale a dedicated deployment up, down or to zero, and delete a file or a GPU cluster.

Reads

• togetherai_list_fine_tunesTogether AI fine-tuning jobs, newest first: status, base model, output model, training type, epochs, tokens and timestamps.
• togetherai_get_fine_tuneOne Together AI fine-tuning job in full, with its most recent events (what stage it reached, warnings and errors).
• togetherai_list_batchesTogether AI batch inference jobs, newest first: status, endpoint, model, progress, input and output file ids, errors and deadlines.
• togetherai_list_endpointsDedicated inference endpoints in the Together AI project, each with its deployments: state, hardware, model, ready and desired replicas…
• togetherai_list_clustersTogether AI GPU clusters: name, type, status, region, GPU type and count, billing type, reservation window and when created.
• togetherai_list_filesFiles uploaded to Together AI by id: purpose (fine-tune, eval, batch-api), type, size, processing status, whether valid and when uploaded.
• togetherai_list_evaluationsTogether AI evaluation jobs: type (classify, score, compare), status, when created and updated, and headline results (pass percentage, mean score…

Acts, when you ask

These change something in Together AI. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• togetherai_create_batchStarts a Together AI batch job that runs every request line in an uploaded file against an endpoint (chat completions…
• togetherai_cancel_batchCancels a Together AI batch job; requests already processed stay billed and the job cannot be resumed.
• togetherai_cancel_fine_tuneCancels a running Together AI fine-tuning job (id starts ft-); training done so far is billed and the job cannot be resumed.
• togetherai_scale_deploymentSets the replica bounds of a dedicated endpoint deployment on Together AI.
• togetherai_delete_filePermanently deletes a file uploaded to Together AI (training, eval or batch data); jobs that need it can no longer use it.
• togetherai_delete_clusterPermanently deletes a Together AI GPU cluster, its nodes and whatever runs on them; billing for it stops.

Connect it

First connect Together AI on Dashboard → Connections, then add Memtro to your client once. The Together AI tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude plugin marketplace add memtro-org/claude-memtro
claude plugin install memtro@memtro --config api_key=sk-memtro-...

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared Together AI connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › Which fine-tuning jobs are running, and what stage has each reached?
• › Did any batch jobs fail this week?
• › How many replicas are ready on each dedicated endpoint?
• › What were the headline results of the latest evaluation?
• › Cancel the batch job that started this morning.

Questions

Is this an official Together AI MCP server?

No. It is Memtro's MCP server with Memtro's Together AI connector behind it: the connector reads Together AI through its API with the credentials you connect on Dashboard → Connections, and Together AI publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in Together AI?

Only when you ask, and only what it has been granted: start a batch inference job from an uploaded file (billed per token); cancel a batch job; cancel a fine-tuning job; scale a dedicated endpoint deployment, or stop it with 0 replicas; delete an uploaded file; delete a GPU cluster. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Engineering and operations

• GitHub MCP server
• AWS CloudWatch MCP server
• Sentry MCP server
• ServiceNow MCP server
• Okta MCP server
• Twilio MCP server