Retrieve metadata for any Replicate open-source AI model via MPP-paid API call.
What it does
The Replicate — Get Model endpoint, hosted through the Locus MPP gateway, returns metadata about a specific model available on Replicate's platform. You supply a model owner (namespace) and model name, and the endpoint returns information about that model. This is useful for programmatic discovery of model details before running inference.
This endpoint is part of a broader Replicate MPP service that also includes endpoints for running models, retrieving predictions, and listing available models. Payment is handled via the MPP (Micropayment Protocol) using the Tempo method, settling in pathUSD on Tempo L2. The Get Model endpoint costs 4,000 base units (pathUSD uses 6 decimals), which works out to $0.004 per call.
The endpoint accepts POST requests with a JSON body containing `model_owner` and `model_name` fields. For example, to look up the Flux Schnell image generation model, you would send `model_owner: "black-forest-labs"` and `model_name: "flux-schnell"`. The response schema is not documented in the OpenAPI spec, but it presumably mirrors Replicate's native model metadata response. Note that the probe returned 404 on HEAD/GET — this endpoint only responds to POST requests, which is consistent with the OpenAPI definition.
Capabilities
Use cases
- —Look up metadata for a specific Replicate model before running inference
- —Verify a model exists and check its latest version programmatically
- —Build agent workflows that discover model capabilities before selecting which to run
- —Integrate Replicate model information into dashboards or catalogs
Fit
Best for
- —AI agents that need to inspect model details before committing to a run
- —Developers building model selection pipelines across Replicate's catalog
- —Automated systems that validate model availability before dispatching jobs
Not for
- —Actually running model inference (use the /replicate/run endpoint instead)
- —Browsing models interactively (use replicate.com directly)
- —Free or unauthenticated access — every call requires MPP payment
Quick start
curl -X POST https://replicate.mpp.paywithlocus.com/replicate/get-model \
-H "Content-Type: application/json" \
-d '{"model_owner": "black-forest-labs", "model_name": "flux-schnell"}'Example
Request
{
"model_name": "flux-schnell",
"model_owner": "black-forest-labs"
}Endpoint
Quality
The OpenAPI spec provides clear request schema, pricing, and endpoint structure. However, the response schema is undocumented, no example responses are available, the probe returned 404 because it used HEAD/GET rather than POST (the only defined method), and crawled pages returned no useful documentation. Pricing is clear from the spec.
Warnings
- —Probe returned 404 on HEAD and GET — endpoint only accepts POST per the OpenAPI spec, so liveness could not be confirmed via the probe
- —Response schema is not documented; actual response structure is unknown
- —No crawled documentation was available — all auxiliary pages returned 'not found' errors
- —pathUSD currency address (0x20c0…b9537d11c60e8b50) could not be independently verified
Citations
- —Get Model endpoint costs 4000 base units in pathUSD via Tempo settlementhttps://replicate.mpp.paywithlocus.com
- —Endpoint accepts POST with model_owner and model_name fieldshttps://replicate.mpp.paywithlocus.com
- —Replicate API reference available at replicate.com/docs/reference/httphttps://replicate.com/docs/reference/http
- —LLM-oriented docs available at beta.paywithlocus.com/mpp/replicate.mdhttps://beta.paywithlocus.com/mpp/replicate.md