LIVE GPU PRICING & SPECS: Compare GPU cloud providers with practical, transparent guides.
Zero Affiliate Markup / Updated Continuously / 🌐 39+ Global Datacenters
Home / Developer API & MCP
REST API v1.0 & MCP 2024-11-05 Sub-10ms API Latency

Query Real-Time GPU Pricing & Hardware Benchmarks via API.

Connect Claude Desktop, Cursor IDE, agentic reasoning models, or internal procurement pipelines directly into the Cloud GPU Rating benchmark and spot price matrix.

12+ Verified AI Clouds
Sub-5ms Median Query Latency
100% MCP 2.0 Compliant
bash — curl /api/v1/pricing 
# Query normalized GPU price matrix across H100, RTX 4090, MI300X
curl -s https://cloudgpurating.com/api/v1/pricing \
  -H "Authorization: Bearer cgr_live_YOUR_SECRET_KEY"

# Response (HTTP 200 OK)
{
  "status": "success",
  "summary": {
    "h100_sxm5_hourly": { "min": 1.99, "median": 2.89, "max": 4.48 },
    "rtx_4090_hourly":  { "min": 0.34, "median": 0.44, "max": 0.79 }
  }
}

🔑 Authentication & API Key Management

All requests to the Cloud GPU Rating REST API and MCP Server require authentication using an active API key prefixed with cgr_live_. API keys can be generated, labeled, rotated, or revoked instantly in the Admin Control Center.

Passing Your API Key

We recommend transmitting your API key via the standard HTTP Authorization header using the Bearer scheme:

Authorization: Bearer cgr_live_YOUR_SECRET_TOKEN

Alternatively, the API supports the custom header X-API-Key: cgr_live_... or query parameter ?api_key=cgr_live_... for rapid curl testing.

⚡ Rate Limits & Response Headers

Rate limits are enforced on a per-key sliding window of 60 seconds. Each response returns informative HTTP headers indicating your current quota:

Header Description Example
X-RateLimit-Limit Maximum allowed requests per minute for this key. 120
X-RateLimit-Remaining Number of requests remaining in current window. 118
X-Response-Time Backend execution and database query latency. 2.14ms
Retry-After Seconds to wait before retrying (returned on HTTP 429). 60

🛡️ Permission Scopes

Keys can be constrained to granular permissions to protect administrative actions:

Scope Access Level Allowed Operations
read:providers Read Only Query provider profiles, categories, and availability filters.
read:pricing Read Only Access aggregate spot rates, H100 hourly pricing, and historic ranges.
read:benchmarks Read Only Access vLLM tokens/sec, InfiniBand latency, and FLOPS per dollar metrics.
read:guides Read Only Query technical guides, key takeaways, and deployment blueprints.
read:ads Read Only Query ad spaces, active sponsor campaigns, and performance impressions/clicks.
write:ads Write Only Create, update, toggle, or upload creatives for ad space campaigns.
write:providers Write Only Create and update provider profiles via POST and PUT endpoints.
mcp:tools Model Protocol Execute JSON-RPC tools and access read-only resources over MCP.
admin:full Superadmin Full access including record deletion and administrative control.
GET /api/v1/health Public (No Key Required)

Returns service uptime, SQLite database connectivity status, and server latency.

curl -s https://cloudgpurating.com/api/v1/health
GET /api/v1/providers Scope: read:providers

Query verified GPU cloud providers with optional filtering parameters:

  • q: Keyword search (e.g. ?q=Runpod or ?q=InfiniBand)
  • cat: Filter by category (e.g. ?cat=Specialized GPU Cloud)
  • gpu: Filter by hardware model (e.g. ?gpu=H100 or ?gpu=RTX+4090)
  • max_price: Maximum starting rate (e.g. ?max_price=1.00)
  • limit & offset: Pagination parameters (default: 50, max: 100)
curl -H "Authorization: Bearer YOUR_KEY" \ "https://cloudgpurating.com/api/v1/providers?gpu=H100&max_price=3.00"
GET /api/v1/pricing Scope: read:pricing

Returns the normalized spot and on-demand GPU pricing matrix with calculated market statistics (min, median, max) across flagship architectures (H100 SXM5, RTX 4090, AMD MI300X, L40S).

curl -H "Authorization: Bearer YOUR_KEY" \ https://cloudgpurating.com/api/v1/pricing
GET /api/v1/benchmarks Scope: read:benchmarks

Access empirical bare-metal interconnect benchmarks (InfiniBand Quantum-2 vs RoCEv2), inference serving throughput across vLLM and TensorRT-LLM (tokens/sec), and FLOPS per dollar calculations.

curl -H "Authorization: Bearer YOUR_KEY" \ https://cloudgpurating.com/api/v1/benchmarks
GET POST PUT /api/v1/ads Scope: read:ads / write:ads

📢 Ad Spaces & Sponsorship Management API

Programmatically manage image banners, HTML/JS code embeds, and native sponsor cards across all website ad spaces.

1. List Available Ad Spaces & Live Capacity

curl -s -H "Authorization: Bearer cgr_live_YOUR_KEY" \ https://cloudgpurating.com/api/v1/ads/slots

2. Query Active Campaigns

curl -s -H "Authorization: Bearer cgr_live_YOUR_KEY" \ "https://cloudgpurating.com/api/v1/ads?slot=header_top&status=active"

3. Create or Deploy New Ad Campaign

curl -s -X POST https://cloudgpurating.com/api/v1/ads \ -H "Authorization: Bearer cgr_live_YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{ "title": "RunPod H100 GPU Cloud Deal", "slot": "header_top", "format": "image", "size_preset": "responsive", "image_url": "https://cloudgpurating.com/assets/providers/runpod.svg", "target_url": "https://runpod.io/?ref=cloudgpurating", "alt_text": "RunPod On-Demand H100 SXM5", "active": 1, "priority": 20 }'

4. Update or Delete Ad Campaign

# Update campaign status curl -s -X PUT https://cloudgpurating.com/api/v1/ads/1 \ -H "Authorization: Bearer cgr_live_YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{"active": 0, "priority": 15}' # Delete campaign curl -s -X DELETE https://cloudgpurating.com/api/v1/ads/1 \ -H "Authorization: Bearer cgr_live_YOUR_KEY"
POST /api/mcp-server.php Official MCP 2024-11-05

🤖 Model Context Protocol (MCP) Integration

Cloud GPU Rating provides a fully compliant Model Context Protocol (MCP) JSON-RPC 2.0 server. This allows AI assistants like Claude Desktop, Cursor IDE, and custom agentic frameworks to invoke GPU cloud lookup tools and read live catalogs directly within conversation context.

Claude Desktop Configuration

Add the following configuration to your claude_desktop_config.json:

{ "mcpServers": { "cloudgpurating": { "command": "curl", "args": [ "-s", "-X", "POST", "https://cloudgpurating.com/api/mcp-server.php", "-H", "Authorization: Bearer cgr_live_YOUR_SECRET_KEY", "-H", "Content-Type: application/json", "-d", "@-" ] } } }

Cursor IDE Configuration

In Cursor, open Cursor Settings > Features > MCP and add a new server or configure .cursor/mcp.json:

{ "mcpServers": { "cloudgpurating": { "url": "https://cloudgpurating.com/api/mcp-server.php", "headers": { "Authorization": "Bearer cgr_live_YOUR_SECRET_KEY" } } } }

Available MCP Tools

Tool Name Purpose Parameters
list_gpu_cloud_providers Filter and search GPU cloud directory query, category, gpu_model, max_price
get_provider_details Fetch full hardware and review profile slug (required)
compare_gpu_pricing Compare hourly rates across clouds gpu_architecture (h100, rtx4090, mi300x)
find_best_gpu_cloud AI workload recommendation engine workload, budget_per_hour, interconnect_required
list_gpu_cloud_guides Query architecture & vLLM blueprints None
list_ads Query ad campaigns by slot and status with live CTR stats slot, status
manage_ad Create, update, toggle active status, or delete ad campaigns action, ad_id, title, slot, format, image_url, target_url
get_ad_spaces Query available ad slots, recommended dimensions, and active count None
list_partners List footer partner badges, directory links, and backlink widgets status: "all"|"active"|"paused"
manage_partner Create, update, toggle active status, or delete footer partner badges action, partner_id, name, format, html_code, ...

Available MCP Resources

  • cgpu://providers/catalog — Complete JSON catalog of verified providers
  • cgpu://pricing/matrix — Normalized hourly pricing dataset
  • cgpu://benchmarks/summary — Interconnect latency and tokens/sec benchmark data

🤖 AI Agent Discovery & LLM Protocols

Cloud GPU Rating provides first-class, machine-readable specifications designed specifically for autonomous LLM crawlers, AI coding assistants (Claude Desktop, Cursor, Copilot, ChatGPT), and retrieval-augmented generation (RAG) pipelines.

Protocol / Standard URL Format Target AI Consumers
LLMs Discovery Standard /llms.txt Markdown Perplexity, Cursor, Copilot, NotebookLM
LLMs Full Knowledge Base /llms-full.txt Extended Markdown Large-context models (Claude 3.5 Sonnet, GPT-4o, Gemini 1.5 Pro)
AI Permissions Standard /ai.txt Text Manifest Autonomous Web Scrapers & Indexers
ChatGPT Plugin Manifest /.well-known/ai-plugin.json JSON OpenAI Custom GPTs & Assistants
OpenAPI 3.1.0 Specification /openapi.json | /openapi.yaml JSON / YAML Swagger, Postman, AutoGPT, LangChain
Model Context Protocol Manifest /.well-known/mcp.json JSON Anthropic Claude Desktop & Cursor IDE
High-Density AI Index /api/v1/ai-index | /ai-index.json Clean JSON Zero-auth RAG pipelines & LLM prompt injection

RAG Ingestion Snippet

Autonomous agents and python crawlers can fetch our live high-density AI index without authentication headers:

curl -s https://cloudgpurating.com/api/v1/ai-index | jq '.market_summary'