Cloud Services
rauta.ai
Manages Rauta gateway accounts and provides GPU availability and model compatibility lookup for AI workloads.
ENDPOINT 1
https://gateway.rauta.ai/mcp
MCP server metadata
- Name
- rauta-gateway
- Version
- 0.1.0
Known tools 15
identify_rautaIdentify this Rauta MCP backend before diagnosing auth or scope errors.
Inferred read-onlyverify_signupSTEP 2 OF 2: complete signup with the 6-digit code the user received by email.
Potential side effectslist_gpusList GPU types currently available across providers, with per-provider price, region, and availability.
Inferred read-onlylookup_modelLook up a Hugging Face model: parameter count, context window, VRAM needed at fp16/int8/int4, and which market GPUs can fit it.
Inferred read-onlyleaderboardInference leaderboard for one category: 'cloud' (GPU rental providers), 'hardware' (raw GPU types), or 'inference' (serverless inference APIs).
Inferred read-onlycold_start_statsReal provisioning (cold-start) latency AND supply health per provider+GPU: p50/p95/min provision time, success/error counts, failure rate, and — for every failing bucket — the actionable cause.
Inferred read-onlycompare_providersCross-provider time series of TTFT p50/p95, latency, tokens/sec and availability.
Inferred read-onlyask_advisorAsk the Rauta Advisor for a single opinionated, prose recommendation grounded in the same live data.
Inferred read-onlyhardware_catalogLook up purchase / ownership prices for AI accelerators (H100, B200, GB200 NVL72, MI300X, etc.).
Potential side effectsprocurement_channelsList buy-side procurement entities (OEMs, distributors, VARs, brokers, colo, etc.) with how_to_buy and buy_url.
Inferred read-onlybuy_vs_reserveRecommend buy (OEM) vs reserve (neocloud) given utilization, horizon, GPU class, and datacenter readiness.
Inferred read-onlysend_feedbackReport a bug or something broken about Rauta WITHOUT AN ACCOUNT -- including when signup or setup is itself the thing that is broken, which is the case this exists for.
Inferred read-onlyCONNECT WITH APPROVAL
Client installation
Review this server and its permissions before adding it. Secret placeholders must be set locally.
Codex
~/.codex/config.toml
[mcp_servers.rauta-gateway]
url = "https://gateway.rauta.ai/mcp"
enabled = true
Claude Code
.mcp.json
{
"mcpServers": {
"rauta-gateway": {
"type": "http",
"url": "https://gateway.rauta.ai/mcp"
}
}
}
Claude Desktop
Settings → Connectors → Add custom connector
Name: rauta-gateway
Remote MCP URL: https://gateway.rauta.ai/mcp
Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.
Cursor
.cursor/mcp.json
{
"mcpServers": {
"rauta-gateway": {
"url": "https://gateway.rauta.ai/mcp"
}
}
}
Visual Studio Code
.vscode/mcp.json
Add to Visual Studio Code{
"servers": {
"rauta-gateway": {
"type": "http",
"url": "https://gateway.rauta.ai/mcp"
}
}
}
Generic MCP
Client-specific MCP configuration
{
"name": "rauta-gateway",
"transport": "streamable-http",
"url": "https://gateway.rauta.ai/mcp"
}
MCP Inspector
Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.
ENDPOINT 2
https://mcp.rauta.ai/mcp
MCP server metadata
- Name
- rauta-market
- Version
- 0.0.1
Rauta tools return real, benchmarker-sourced GPU market + provisioning data. Never invent prices, latencies, or availability — call a tool. Surface real failure rates (e.g. a provider at 0% provision success) rather than hiding providers. Use MCP hosts mcp.rauta.ai / gateway.rauta.ai only (legacy *.infra-arena.ai hosts detach 2026-08-10).
Known tools 10
list_gpusList GPU types currently available across providers, with per-provider price, region, and availability.
Inferred read-onlylookup_modelLook up a Hugging Face model: parameter count, context window, VRAM needed at fp16/int8/int4, and which market GPUs can fit it.
Inferred read-onlyleaderboardInference leaderboard for one category: 'cloud' (GPU rental providers), 'hardware' (raw GPU types), or 'inference' (serverless inference APIs).
Inferred read-onlycold_start_statsReal provisioning (cold-start) latency AND supply health per provider+GPU: p50/p95/min provision time, success/error counts, failure rate, and — for every failing bucket — the actionable cause.
Inferred read-onlycompare_providersCross-provider time series of TTFT p50/p95, latency, tokens/sec and availability.
Inferred read-onlyask_advisorAsk the Rauta Advisor for a single opinionated, prose recommendation grounded in the same live data.
Inferred read-onlyhardware_catalogLook up purchase / ownership prices for AI accelerators (H100, B200, GB200 NVL72, MI300X, etc.).
Potential side effectsprocurement_channelsList buy-side procurement entities (OEMs, distributors, VARs, brokers, colo, etc.) with how_to_buy and buy_url.
Inferred read-onlybuy_vs_reserveRecommend buy (OEM) vs reserve (neocloud) given utilization, horizon, GPU class, and datacenter readiness.
Inferred read-onlyCONNECT WITH APPROVAL
Client installation
Review this server and its permissions before adding it. Secret placeholders must be set locally.
Codex
~/.codex/config.toml
[mcp_servers.rauta-market]
url = "https://mcp.rauta.ai/mcp"
enabled = true
Claude Code
.mcp.json
{
"mcpServers": {
"rauta-market": {
"type": "http",
"url": "https://mcp.rauta.ai/mcp"
}
}
}
Claude Desktop
Settings → Connectors → Add custom connector
Name: rauta-market
Remote MCP URL: https://mcp.rauta.ai/mcp
Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.
Cursor
.cursor/mcp.json
{
"mcpServers": {
"rauta-market": {
"url": "https://mcp.rauta.ai/mcp"
}
}
}
Visual Studio Code
.vscode/mcp.json
Add to Visual Studio Code{
"servers": {
"rauta-market": {
"type": "http",
"url": "https://mcp.rauta.ai/mcp"
}
}
}
Generic MCP
Client-specific MCP configuration
{
"name": "rauta-market",
"transport": "streamable-http",
"url": "https://mcp.rauta.ai/mcp"
}
MCP Inspector
Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.
TRUST AND VERIFICATION EVIDENCE
Loading Trust v2 evidence…
Checking the associated registrable domain. The BuiltWith key remains server-side.
Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.