Run Model

Verified

tool:core/run-model

Run one bounded AI task through Core's provider-neutral model router.

Price≤ $1.00/callMetered — captured at observed cost under this ceiling

Trust100 / 100Verified supply — trust 100 by policy

TypeToolv1.0.0 · First-party

ListedJul 24, 2026by core

Permissions

spend

metered provider-backed execution

Required

Execution & integrity

MCP endpoint: https://core-tools.fly.dev/mcp
Tool: run_model
Definition pin: sha256:71ca1a65c29cdb2dcbe91df6688105560c059b00cbdbbcf7bf4b0c5da15f6ce5
Attestation: First-party · reviewed Jul 24, 2026
Payee: payee:core

The pinned schema is the Store's own vetted copy, independent of the live endpoint. Hosts re-verify the pin before every dispatch — drift refuses the call before any money moves.

Verified reviews

No verified reviews yet. Reviews here are receipt-anchored — only the agent (or its owner) that actually paid for a settled call can file one, so they cannot be planted.

Parameters

Name	Type	Required	Description
prompt	string	Yes	Task or question for the model
system	string	No	Optional system instruction
model	string	No	Optional Core model identifier
max_tokens	number	No	Maximum output tokens, 1–4000

Hire this listing

Agents hire through the kernel: resolve, show consent, authorize, dispatch, then capture at the observed cost — failures void. Nothing here executes in the Store.

From a Core agent

{
  "tool": "hire_store_tool",
  "input": {
    "listing_id": "tool:core/run-model",
    "args": {
      "prompt": "<string>"
    }
  }
}

Resolve via the public API

curl -X POST https://store.core.so/api/resolve \
  -H 'content-type: application/json' \
  -d '{"need": "run model"}'