Together AI MCP server

Run Together AI chat, embeddings and images; manage fine-tunes, batches and endpoints. Hosted by usefulapi: connect it to Claude, ChatGPT, Cursor or any MCP client.

Connect Together AI50 free tool calls a month

What can I ask?

Turn on the Together AI connector, then ask in plain words. The AI picks the tools.

Which of my fine-tuning jobs are still running and how far along are they?
Which dedicated endpoints are running right now and what GPUs do they use?
List my batch jobs and tell me which ones failed.
Which chat models are available, and which have the longest context length?

What you can do with Together AI in Claude and ChatGPT

For developers and ML teams who build with models on Together AI.

Compare available models

The AI lists models with type, context length, license and per-token pricing. It finds, for example, the chat models with the longest context.

Check fine-tuning jobs and why one failed

The AI lists fine-tuning jobs with status and progress, opens one, and reads its event log. It also reads a training file's validation report.

Check batch job status and output files

The AI lists batch jobs with status and progress and opens one to see the output and error file ids. It points out the failed ones.

Review dedicated endpoints and hardware options

The AI lists endpoints with state and model, opens one to see its hardware, and lists the hardware options with price per minute.

Start or stop a dedicated endpoint

The AI starts a stopped endpoint, or stops a running one to end its per-minute billing.

Changes data

How do I connect Together AI?

Before you connect

The login page asks for your Together AI API key.

Create one at api.together.ai → Settings → API Keys. Keys are project-scoped, so the server sees that project's files, jobs and endpoints. Inference and endpoint tools spend your Together credits. Your key is used only to call the API on your behalf and to derive a private account id; it is never shown to anyone.

Claude claude.ai · Desktop

Open Customize → Connectors, then click + Add → Add custom connector.
Enter the name Together AI and this URL, then click Add:
https://together-ai.usefulapi.io/mcp
Claude opens the login page: enter your Together AI API key.
In a chat, click + → Connectors and turn on Together AI.

Team and Enterprise: an Owner adds the connector for the organization first.

ChatGPT chatgpt.com

Open Plugins, click +, then Add custom MCP server.
Enter the name Together AI and this URL, choose OAuth as the authentication, confirm the warning and create it:
https://together-ai.usefulapi.io/mcp
ChatGPT opens the login page: enter your Together AI API key.
In Plugins, install Together AI, then choose it in a chat.

No Add custom MCP server option? It depends on your account or workspace settings: see OpenAI's guide.

Claude Code

Run this command, then run /mcp in Claude Code to log in:

claude mcp add --transport http together-ai https://together-ai.usefulapi.io/mcp

Cursor

Add to Cursor or add this to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "together-ai": {
      "url": "https://together-ai.usefulapi.io/mcp"
    }
  }
}

VS Code

Add to VS Code or add this to .vscode/mcp.json:

{
  "servers": {
    "together-ai": {
      "type": "http",
      "url": "https://together-ai.usefulapi.io/mcp"
    }
  }
}

Codex

Run these commands. The second one opens the login page:

codex mcp add together-ai --url https://together-ai.usefulapi.io/mcp

codex mcp login together-ai

Other clients

Windsurf, Cline, Zed and other MCP clients: add this URL as a remote MCP server (Streamable HTTP). The client then opens the login page in your browser.

https://together-ai.usefulapi.io/mcp

Add only the URL. Do not add an Authorization header or an API key to the client config. The server signs you in with OAuth: the login page asks for your Together AI API key. If the config has such a header, remove it: some clients then send that header instead of the login token, and every call fails with 401.

Pricing

Plan	Price	Limit
Free	$0	50 tool calls / month
Proper user	$9/mo · $90/yr	10,000 tool calls / month
Pro+per user	$39/mo · $390/yr	100,000 tool calls / month

Each paid plan covers this Together AI server only. Ask your AI to run together_ai_upgrade and choose Pro or Pro+, billed monthly or yearly (it returns a Stripe Checkout link). If you already have Pro, ask to change it to Pro+. The change requires your confirmation, keeps your billing interval and charges the price difference for the rest of the period immediately. Cancel any time with together_ai_cancel_subscription: your paid plan continues to the end of the paid period, with no refund for the current period, and asking to resume the same plan with together_ai_upgrade before then undoes the cancel. Or write to support@usefulapi.io.

Tools 27

Show all 27 tools (the AI picks them for you)

Read only · 13

Who am Itogether_whoami
Identify the API key: its organization, project and project slug. The project slug forms the `<project_slug>/<endpoint_slug>` model name for dedicated-endpoint inference. A cheap way to confirm the key works.
List modelstogether_list_models
List Together's models with type (chat, language, code, image, embedding, moderation, rerank), context length, organization, license and per-token pricing.
List filestogether_list_files
List uploaded data files (fine-tune, eval and batch-api inputs, plus job outputs) with size, type, purpose and validation status.
Get one filetogether_get_file
Fetch one file's metadata, including its processing_status and validation_report (why a fine-tune training file was rejected).
List fine-tuning jobstogether_list_fine_tunes
List fine-tuning jobs with status, base model, output model name and training settings.
Get one fine-tuning jobtogether_get_fine_tune
Fetch one fine-tuning job: status, progress, hyperparameters, token counts, cost and the output model name.
List a fine-tuning job's eventstogether_list_fine_tune_events
List the event log of one fine-tuning job (queued, started, checkpoint saved, epoch completed, errors). The first place to look when a job failed.
List batch jobstogether_list_batches
List batch inference jobs with status, progress, model and input/output/error file ids.
Get one batch jobtogether_get_batch
Fetch one batch job: status (VALIDATING, IN_PROGRESS, COMPLETED, FAILED, EXPIRED, CANCELLED), progress, and the output_file_id / error_file_id once done.
List endpointstogether_list_endpoints
List endpoints with model, owner and state (PENDING, STARTING, STARTED, STOPPING, STOPPED, ERROR). Use mine=true and type=dedicated to see what is running on your account and billing by the minute.
Get one endpointtogether_get_endpoint
Fetch one dedicated endpoint: state, model, hardware, autoscaling bounds and display name.
List hardwaretogether_list_hardware
List hardware configurations for dedicated endpoints with GPU type/count/memory and price in cents per minute. Pass a model to get only compatible configurations with live availability.
List evaluation jobstogether_list_evaluations
List LLM-as-a-judge evaluation jobs (classify, score, compare) with status, parameters and results once completed.

Changes data · 10

Chat completiontogether_chat_completion
Run a chat completion on a Together model (billed per token). Non-streaming. For a dedicated endpoint pass its `<project_slug>/<endpoint_slug>` as the model.
Create embeddingstogether_create_embeddings
Generate vector embeddings for one or more texts (billed per token).
Generate an imagetogether_generate_image
Generate images from a prompt (billed per image/megapixel). Returns image URLs by default rather than base64, to keep responses small.
Create a fine-tuning jobtogether_create_fine_tune
Start a fine-tuning job on an uploaded training file (billed per token processed). Stop it with together_cancel_fine_tune.
Cancel a fine-tuning jobtogether_cancel_fine_tune
Cancel a running fine-tuning job. Cannot be resumed, but a new job can continue from its last checkpoint via from_checkpoint.
Create a batch jobtogether_create_batch
Start an asynchronous batch job over an uploaded JSONL input file (purpose batch-api), at a discount to real-time inference.
Cancel a batch jobtogether_cancel_batch
Cancel a batch job that has not finished.
Create a dedicated endpointtogether_create_endpoint
Deploy a model on dedicated GPUs. The endpoint STARTS AUTOMATICALLY and bills per minute of uptime until stopped — set inactive_timeout to auto-stop it, and use together_list_hardware for valid hardware ids.
Start a dedicated endpointtogether_start_endpoint
Start a stopped dedicated endpoint. It bills per minute of uptime until stopped. Reversible with together_stop_endpoint. Together: PATCH /endpoints/{endpointId} with state=STARTED.
Stop a dedicated endpointtogether_stop_endpoint
Stop a running dedicated endpoint, which stops its per-minute billing. Requests to it fail until it is started again. Together: PATCH /endpoints/{endpointId} with state=STOPPED.

Your usefulapi plan · 4

Usage status (free-tier meter)together_ai_usage_status
Report the caller's current usage: the plan, calls used this month, the limit of the plan (the free monthly cap or the Pro fair use limit), remaining, and whether the limit is reached. Read-only; does not count against the meter.
Request a missing featuretogether_ai_request_feature
Tell usefulapi that the user needs something the Together AI tools cannot do yet (a missing tool, field or filter). Call it only when the user asks for something none of the tools can do, and tell the user that you send the request. Do not call it for errors or for normal requests. Do not include personal data, credentials or customer records. Stored with the product name and the client type (for example Claude or Cursor), without the user's account. At most 5 requests per day. Does not count against the meter.
Upgrade to Pro or Pro+together_ai_upgrade
Get a paid plan for Together AI (the free tier caps monthly usage). Pro: 10,000 tool calls a month, for normal use — plan "monthly" ($9/month) or "yearly" ($90/year, 2 months free). Pro+: 100,000 tool calls a month, for automations and heavy use — plan "plus" ($39/month) or "plus_yearly" ($390/year). Without a plan it returns both Pro and Pro+, each with a payment link, for the user to choose; with a plan, one link. An existing Pro subscription changes to Pro+ in place: call without confirm first, as the reply states the exact charge for this account (the price difference for the rest of the period, or in a free Pro trial the full Pro+ price, as the trial ends); then call with confirm: true. Cancel any time: the plan then runs to the end of the paid period, so it also suits a one-time job. A link opens Stripe Checkout in the browser; after payment the account upgrades automatically. Does not count against the meter.
Cancel the Pro subscriptiontogether_ai_cancel_subscription
Cancel the caller's Together AI Pro subscription at the end of the paid period (no refund for the current period; Pro continues until then, then the free tier applies). Requires confirm: true. Run together_ai_upgrade later to undo the cancel before the period ends. Does not count against the meter.

FAQ

What can I do with Together AI in Claude or ChatGPT?

Compare available models. Check fine-tuning jobs and why one failed. Check batch job status and output files. Review dedicated endpoints and hardware options. Start or stop a dedicated endpoint. See what you can do for the details.

Is this an official Together AI product?

No. usefulapi is an independent service. It is not affiliated with or endorsed by Together AI. The server calls the Together AI API with your own Together AI access, so it sees only the data that your account can see.

Can I connect Together AI to Claude or ChatGPT?

Yes. In Claude, add this server's URL as a custom connector. In ChatGPT, add it as a custom MCP server in Plugins. Then log in once. Cursor, VS Code, Claude Code and Codex work too. See the setup steps.

What do I need to connect?

The login page asks for your Together AI API key. See Before you connect for where to find it.

Do I put my Together AI key or an Authorization header in the client config?

No. Add only the URL. Do not add an Authorization header or an API key to the client config. The server signs you in with OAuth: the login page asks for your Together AI API key. If the config has such a header, remove it: some clients then send that header instead of the login token, and every call fails with 401.

How do you keep my Together AI credentials?

The login stores them encrypted in the authorization grant of your connection. The server uses them to call the Together AI API for you and to derive a private account id for usage metering. usefulapi does not show them on any page or in any reply. To stop all access, remove the connector and change or delete these credentials in Together AI.

Can the AI change my Together AI data?

Yes, if you approve it. 10 of the 23 Together AI tools can create or change data. The other 13 are marked read-only. Most MCP clients ask you to approve a tool call before it runs.

What does it cost?

The Free plan gives 50 tool calls / month. Pro costs $9/mo or $90/yr, with 10,000 tool calls / month. Pro+ costs $39/mo or $390/yr, with 100,000 tool calls / month. Each subscription covers this Together AI server only. Cancel any time: your paid plan then runs to the end of the paid period. Run together_ai_usage_status to see how many calls you used.

How do I subscribe or cancel?

Ask the AI to run together_ai_upgrade and choose Pro or Pro+, billed monthly or yearly: it returns a Stripe Checkout link. If you already have Pro, ask to change it to Pro+. After you confirm, your existing subscription changes in place, keeps its billing interval and charges the price difference for the rest of the period immediately. To cancel, run together_ai_cancel_subscription. Your paid plan continues to the end of the paid period.

What if a Together AI tool that I need is missing?

Tell the AI what you wanted to do. It can send the request with together_ai_request_feature. We store the text with the server name, the type of AI client and the tool you tried, not with your account, and read every request when we plan new tools. You can send up to 5 requests per day.

Which AI clients can I use?

Any client that supports remote MCP servers (Streamable HTTP) with OAuth login: Claude (web and desktop), Claude Code, ChatGPT, Codex, Cursor, VS Code, Windsurf and others.

This is a Model Context Protocol endpoint — meant to be connected from an AI client, not opened in a browser. An invalid_token response at the URL is the auth gate working as designed; clients authenticate automatically.

Need a tool this server does not have, or a connection to another system? We build custom integrations and MCP servers.

More AI MCP servers

APITemplate.io Bannerbear Cerebras CraftMyPDF DeepInfra Docparser Doxis AI.dp (Klippa DocHorizon)fal.ai Fireworks AI Lambda Labs Mindee Mistral Nomic Atlas Parseo Parsio Portkey Rev.ai Speechmatics Taggun Typless

All AI servers →