- ACTIVE
- SUCCESS
- FAILURE
- HUMAN
Short answer
How can I fine tune models using my AI?
You tell your agent to start a training run on the Together AI cluster. It handles the job creation and monitors the status so you can see when your model is ready.
Training outcomes
Where your training runs land.
The AI manages the cluster workload so you can focus on the data.
ACTIVE
Running jobs
Your AI identifies which training runs are currently in progress. You'll see exactly what's running on the cluster right now.
SUCCESS
Completed models
When a job finishes, your agent confirms the success. You'll know exactly when your custom weights are ready to use.
FAILURE
Error tracking
If a run hits a snag, the AI flags the failed job. You'll get clear visibility into which attempts didn't work.
HUMAN
Manual review
The AI manages the cluster, but you decide if the model is good enough. You retain full control over which weights to deploy.
The workflow
What your AI does when you request a fine tune.
Your agent acts as the bridge between your chat and the Together AI compute cluster.
Start a run
Your agent submits your training parameters and dataset location to the cluster.
create_finetune_jobCheck status
The AI pulls a list of all current jobs to see how they are progressing.
list_finetune_jobsTrack progress
Your agent monitors the job list to find updates on specific training tasks.
list_finetune_jobsConfirm completion
The AI verifies that a job has finished successfully before you try to use the model.
list_finetune_jobs
Try it
Copy these to start.
Use these prompts to interact with the Together AI cluster.
Starting points
Swap out the model names and dataset paths for your own specific files.
Together AI Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_uwxFbBTrWtpKdtNP0Ee38Zmxxir6fie6FBKXtn0o/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Together AI capabilities are ready to use.
{
"mcpServers": {
"together-ai-mcp": {
"url": "https://edge.vinkius.com/vk_preview_uwxFbBTrWtpKdtNP0Ee38Zmxxir6fie6FBKXtn0o/mcp"
}
}
}Start a fine-tuning job for Llama-3 using my dataset at s3://my-bucket/data.jsonl
Show me all the fine-tuning jobs I've run on Together AI recently.
Check the status of my last training job to see if it's finished.
List all failed fine-tuning jobs so I can see what went wrong.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Together AI once, then ask.
Link your account once and your credentials stay encrypted. You can then manage your entire training pipeline from your chat interface.
Connect Together AI to your AIFAQ
How this task behaves
- 01
Can my AI delete a training job?
No. The AI can only create or list jobs. You have to handle deletions through the Together AI dashboard.
- 02
Can the AI upload my dataset files?
No. You need to provide a path to a dataset that is already hosted on a cloud provider.
- 03
Can the AI see my training costs?
No. The AI only has access to job status and creation; it cannot access your billing or usage data.
- 04
Will the AI tell me if a job failed?
Yes. It can check the status of your jobs and report if they have failed.
- 05
Can I change a job while it's running?
No. The AI can only start new jobs or list existing ones. It cannot modify a job once it's in progress.
More questions about Together AI? See everything the Together AI Connector can do
Connect Together AI to Claude, Cursor, ChatGPT & more
