Vast.ai for AI agents

Vast.ai is a marketplace for renting GPU compute, used by teams that need affordable, on-demand hardware for training or running AI models. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.

AI and modelsAPI keyMCP + RESTvast.ai
AUDIT LOG · VAST.AIPOLICY: acme-support
09:41:07 · claude · u_8f2read
vast_ai.search_templates
Search Templates✓ allowed · 212ms
09:41:08 · claude · u_8f2read
vast_ai.search_gpu_offers
Search GPU Offers✓ allowed · 164ms
EVERY VAST.AI CALL, ON THE RECORD
01 · USE CASES

What agents do in Vast.ai.

01

Search GPU offers

Find available GPU instances matching a price and specification, a read-only search.

02

Search launch templates

Look up pre-configured templates for common workloads before renting an instance.

03

Search storage volume offers

Check available storage options to pair with a rented GPU instance.

02 · ACTIONS

3 Vast.ai actions, graded by risk.

Every Vast.ai action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.

read

3

Look things up. Allowed by default.

  • vast_ai.search_templates
    Search Templates
  • vast_ai.search_gpu_offers
    Search GPU Offers
  • vast_ai.search_volume_offers
    Search Volume Offers

write

0

Create and change things. Allow, or ask the user first.

  • No write actions.

destructive

0

Delete, cancel or archive. Ask first, or deny outright.

  • No destructive actions.
03 · HOW IT WORKS

Vast.ai in three steps.

  1. 01Your users connect Vast.aiThey add their Vast.ai api key on Arc0 Connect, under your brand. It goes straight into the vault.
  2. 02You set the rulesReads run, and writes like “updates” can wait for the user to approve.
  3. 03Any agent can actYour agent calls Vast.ai through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
POLICY.TS
await arc0.policies.set('vast_ai', {
  read: 'allow',
  write: 'ask',        
  destructive: 'deny',  
})

# Claude Code: the same connection, one URL
$ claude mcp add --transport http arc0 \
    https://mcp.arc0.ai/u/u_8f2
04 · AUTH AND DATA

How Vast.ai connects.

Users add their Vast.ai api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.

The same Vast.ai connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →

AUTH
API key
CREDENTIALS
Per-tenant encrypted vault
MODEL SEES
Results only, never credentials
AUDIT LOG
Every call, on every plan
05 · WORKS WITH

Use Vast.ai from any agent.

Claude
ChatGPT
Cursor
Codex
VS Code
OpenAI Agents SDK
Claude Agent SDK
Vercel AI SDK
Mastra
LangGraph
07 · FAQ

Vast.ai and Arc0, answered.

Q01

Can I use Vast.ai with Claude, ChatGPT or Cursor?

Yes. Connect Vast.ai to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Vast.ai actions you allow.

Q02

How do users connect Vast.ai?

Users add their Vast.ai api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.

Q03

Which Vast.ai actions can my agent take?

3 in total: 3 read, 0 write and 0 destructive. Your policies decide which of them each agent may call.

Q04

Can I make my agent read-only in Vast.ai?

Yes. Allow read actions and deny writes in the Vast.ai policy. Your agent can still look things up, and any write it attempts is blocked and logged.

Q05

Can my own backend call Vast.ai too?

Yes. The same Vast.ai connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.

Get started

Plug Vast.ai into your agent.

Your users connect Vast.ai once, under your brand. Your agent gets 3 actions behind your policies, with every call on the record.

Free to build · MCP + REST · Audit log on every plan