Public trust snapshot

See what the marketplace can serve right now.

Live, aggregate health for remote GPU inference. We publish enough to verify capacity and reliability without exposing provider identities, node ids, prompts, API keys, or customer data.

Open Agent Connect View public JSON

Routing requirements

Every request passes the same four checks.

A node does not become eligible just because a GPU exists.

Authenticated request

Customer API key, model scope, rate limit, and credits are checked first.

Verified supply

Only approved providers and nodes may receive work.

Fresh and capable

Heartbeat, model, VRAM, runtime, capacity, and worker version must match.

Private boundary

Customer files, tools, browser, and shell stay on the customer computer.

Verified providers

4Approved by platform review

GPU nodes online

0Fresh verified heartbeats

Routable models

0 / 5Available / active catalog

Average latency

No sampleSuccessful requests in 24h

24h successNo sample

Stale nodes0

Tokens served in 24h0

Average trust30.0

Operational health

Requests and anonymous GPU capacity

Succeeded0

Failed / expired0

Latency bandAwaiting sampleA latency band appears after a request completes.

Successful requests0%

24GB+ VRAM0/1 online

12GB VRAM0/4 online

8GB VRAM0/1 online

Trust boundaries

What the platform verifies and never exposes

No fresh capacity

No verified GPU is fresh enough for new routing.

Provider approval is required before scheduling.
Fresh heartbeat, model capability, and capacity are required.
Provider identities and exact node metadata remain private.
Customer tools, files, shell, and browser never move to a provider.

Verify it yourself

Three commands from health check to agent endpoint.

Copy only the values your tool needs.

01 / PUBLICCheck marketplace health

No login or API key required.

curl https://gpuintegrate.com/api/public/trust

02 / MODELSDiscover routable models

Use your customer API key.

curl https://gpuintegrate.com/v1/models -H "Authorization: Bearer gc_live_..."

03 / CONNECTConfigure your local agent

Works with OpenAI-compatible clients.

OPENAI_BASE_URL=https://gpuintegrate.com/v1
OPENAI_API_KEY=gc_live_...
OPENAI_MODEL=llama3.2:3b

Anonymous provider trust

Reliability signals without exposing who owns the GPU.

Verified GPU Provider #10/2 nodes online / 15.9GB VRAM class

Verified

75.0 trust / 0 completed / 0 tokens

Verified GPU Provider #20/2 nodes online / 15.9GB VRAM class

Verified

0.0 trust / 0 completed / 0 tokens

Verified GPU Provider #30/1 nodes online / 24GB VRAM class

Verified

0.0 trust / 0 completed / 0 tokens

Verified GPU Provider #40/1 nodes online / 8GB VRAM class

Verified

0.0 trust / 0 completed / 0 tokens