Together AI

What it does

Safeguard reads an inventory of what your Together AI project runs: its dedicated endpoints (model, hardware, autoscaling and state), its fine-tuning jobs (base model, output model and method), the catalog models those use with the licence Together reports for each, and the organization's month-to-date usage totals. The results appear as AI asset inventory entries and continuous tests (model provenance and licence risk). Find the Together AI card under Integrations on the Project/Product tab (the AI Model filter shows only AI model cards). Nothing it reads is sent to a model: no customer data goes to third-party LLMs.

Public vs private

Supported auth methods

Permissions required

Safeguard only reads: it sends GET requests and nothing else. Together AI documents no read-only key: any key can also run inference and start jobs, so treat it as an elevated grant.

Project API key

Least privilege: one project API key per project, with an expiry date. There is no narrower key.

Legacy API key

The same table applies. A legacy key reads whatever the account can, so prefer a project key.

Token types

Expiry: set an expiration date on the key (for example 90 days). When it expires the connection test fails with a rejected key; create a new one and update the connection.

What we never need: files, chat or completion calls, job creation, or member management. Safeguard never reads training files, prompts or completions.

Step-by-step setup

Create the credential in Together AI

1. Sign in to Together AI and switch to the project to inventory.
2. Open the project's API keys and create a key named Safeguard (read-only use). Set an expiration date if your policy needs one.
3. Copy the key. Together shows a new key only once.

Or with the Together CLI

Create the key in the dashboard; Safeguard's setup has no CLI step for it.

Connect it in Safeguard

1. Open Integrations and choose the Project/Product tab (optionally the AI Model filter).
2. Click Connect on the Together AI card.
3. Give the connection a name and paste the key into the API key field. The Setup steps section repeats the key steps above, with a copy button.
4. Save, then click Check access. Safeguard calls GET /v1/whoami with the key. Success reads The key reads Together AI project <project> of <organization>. Together has no read-only key: it can also run inference and start jobs; Safeguard only reads. Add one connection per project.

Webhooks and triggers

This integration has no webhooks. The inventory is read each time the connection's tests run, on their schedule or when you run them by hand.

What we read and store

• Read: the key's project and organization, the project's dedicated endpoints, its fine-tuning jobs, the catalog entries of the models they use, and the month's usage line items.
• Stored: the inventory (endpoint and model names, hardware, autoscaling, job status, base and output models, licences) and, from usage, only the month's total cost, the number of line items and the product names, in the AI asset inventory, and the test results. An asset that disappears is marked inactive, not deleted. The key is encrypted at rest and never shown again after you save it.
• Not stored or read: training and validation files, job events, prompts and completions.
• No customer data from this integration is sent to third-party LLMs.

Troubleshooting

Rotation and disconnect

• Rotate: create a new key in the project, edit the connection in Safeguard, paste it and save, then revoke the old key.
• Disconnect: delete the connection in Safeguard, then revoke the key in Together AI.

Vendor references

• Together AI API keys and authentication, checked 2026-10-16.
• Who am I, checked 2026-10-16.
• List endpoints, checked 2026-10-16.
• Get endpoint, checked 2026-10-16.
• List fine-tunes, checked 2026-10-16.
• List models, checked 2026-10-16.
• Billing usage, checked 2026-10-16.