What is This Skill?
This skill teaches AI coding agents how to provision and manage GPU clusters on Together AI. Request on-demand or reserved clusters of H100, H200, and B200 GPUs for large-scale model training, distributed inference, and research workloads.
Answer-Ready: Together AI GPU Clusters Skill for coding agents. Provision H100/H200/B200 GPU clusters on-demand or reserved. Large-scale training and distributed inference. Part of official 12-skill collection.
Best for: Teams needing GPU clusters for training or large-scale inference. Works with: Claude Code, Cursor, Codex CLI.
What the Agent Learns
Provision Cluster
from together import Together
client = Together()
cluster = client.clusters.create(
name="training-cluster",
gpu_type="h100-80gb",
gpu_count=8,
reservation_type="on-demand",
)
print(f"Cluster ID: {cluster.id}")GPU Options
| GPU | VRAM | Interconnect | Best For |
|---|---|---|---|
| H100 80GB | 80GB | NVLink | Standard training |
| H200 | 141GB | NVLink | Large models |
| B200 | 192GB | NVLink | Cutting-edge |
Reservation Types
| Type | Billing | Commitment |
|---|---|---|
| On-demand | Per hour | None |
| Reserved | Discounted | 1-12 months |
Cluster Management
# Monitor utilization
status = client.clusters.retrieve(cluster.id)
# Resize
client.clusters.update(cluster.id, gpu_count=16)
# Release
client.clusters.delete(cluster.id)FAQ
Q: How many GPUs can I request? A: From single GPUs to clusters of 1000+ for large training runs. Contact Together AI for very large allocations.