Search GPUs on Vast.ai

VerifiedSafe

Searches for available GPU machines on Vast.ai based on specific requirements like GPU model, VRAM, budget, and region. Helps compare offers and select the best machine for renting GPUs.

Sby Skills Guide Bot
Data & AIIntermediate
606/2/2026
Claude Code
#gpu#vast-ai#cloud-computing#machine-learning#pricing

Recommended for

Our review

Searches for available GPU machines on Vast.ai to rent, matching specific requirements like GPU model, VRAM, budget, and region.

Strengths

  • Access to a wide inventory of GPUs (RTX 4090, A100, H100, etc.)
  • Precise filters (VRAM, GPU count, price, reliability, location)
  • Quick comparison of on-demand, spot, and reserved pricing

Limitations

  • Requires knowledge of Vast.ai-specific query parameters
  • Results depend on real-time offer availability
  • Does not handle instance reservation or launch
When to use it

When you need to quickly find the best rental GPU for a machine learning or compute-intensive project.

When not to use it

When you are looking to purchase GPU hardware or use managed cloud services like AWS SageMaker.

Security analysis

Safe
Quality score90/100

The skill uses Bash only to run vastai search offers with a query constructed from a predefined mapping of user requirements to query fields. No raw user input is directly injected into shell commands, and no destructive or exfiltrating actions are instructed.

No concerns found

Examples

Cheap single RTX 4090
Find a single RTX 4090 GPU for under $1 per hour with good reliability.
8x H100 for large training
Search for 8 H100 GPUs in a datacenter for large model training.
Budget multi-GPU with at least 24GB VRAM
I need at least 2 GPUs with at least 24GB VRAM each, total cost under $1.50/hr.

name: search-gpus description: "Search for available GPU machines on Vast.ai. Use when looking for GPUs to rent, comparing GPU pricing, or finding machines that match specific requirements." argument-hint: "[gpu-type or requirements]" allowed-tools: Bash

Search Vast.ai GPU Offers

Help the user find the best GPU offers on Vast.ai.

User Request

$ARGUMENTS

Instructions

  1. Understand requirements: Key dimensions to determine:

    • GPU model (RTX 4090, A100, H100, etc.)
    • Number of GPUs
    • VRAM requirements
    • Budget ($/hr)
    • Region preferences
    • Reliability needs
    • Pricing type (on-demand vs spot/interruptible vs reserved)
  2. Build the search query from the mapping below.

  3. Run the search and present results clearly.

  4. Help interpret: Highlight best options based on user priorities.

Query Construction

| User says | Query field | |-----------|-------------| | "4090", "RTX 4090" | gpu_name=RTX_4090 | | "A100 80GB" | gpu_name=A100 gpu_ram>=80 | | "H100" | gpu_name=H100 | | "4 GPUs", "multi-GPU" | num_gpus>=4 | | "at least 48GB VRAM" | gpu_total_ram>=48 | | "under $1/hr", "cheap" | dph_total<1.0 | | "reliable" | reliability>0.95 | | "US only" | geolocation=US | | "datacenter" | datacenter=true | | "spot", "interruptible" | add -b flag | | "reserved" | add -r flag | | "NVLink" | bw_nvlink>0 | | "fast internet" | inet_down>500 |

Always include reliability>0.9 unless the user wants cheaper unreliable machines.

Default sort: -o 'dph_total' (cheapest first).

Examples

# Cheap single 4090
vastai search offers 'gpu_name=RTX_4090 num_gpus=1 reliability>0.9' -o 'dph_total'

# 8x H100 for large training
vastai search offers 'gpu_name=H100 num_gpus=8 reliability>0.95' -o 'dph_total'

# Any GPU with 80GB+ VRAM in US datacenters
vastai search offers 'gpu_total_ram>=80 geolocation=US datacenter=true reliability>0.95' -o 'dph_total'

# Budget multi-GPU fine-tuning
vastai search offers 'num_gpus>=2 gpu_ram>=24 dph_total<1.5 reliability>0.9' -o 'dph_total'

# Spot pricing for H100s
vastai search offers 'gpu_name=H100 num_gpus>=4' -b -o 'dph_total'

Presenting Results

After running the search:

  1. Summarize top 3-5 options in a clear table with: Offer ID, GPU, # GPUs, VRAM, $/hr, Reliability, Location
  2. Highlight the offer ID prominently — user needs it to launch
  3. Note trade-offs (price vs reliability vs perf)
  4. Mention spot pricing if it would save significantly
  5. If no results, suggest relaxing constraints
Related skills