List Templates

GEThttps://{organization}.blackbox.ai/dedicated/templates

List every model you can deploy as a dedicated endpoint, along with the valid GPU / region / count combinations for each.

Returns the full catalog of models available for dedicated deployment. Each template exposes the flavors (quantizations / fine-tunes) it ships with, and — for each flavor — the concrete gpu_type, gpu_count, and region combinations you can pass to POST /dedicated/create-deployment.

Call this endpoint first. Deployment creation strictly validates model_name, flavor_name, gpu_type, gpu_count, and region against this catalog — hard-coding values against a stale copy is a common source of 400 invalid_request errors.

Headers

Authorizationstringrequired

API Key of the form Bearer <api_key>.

Query Parameters

typestring

Filter templates by model type.

text2text — chat / completion models
embedding — embedding models

modelstring

Case-insensitive substring match against template name. Useful for narrowing large catalogs (e.g. ?model=gpt-oss matches every openai/gpt-oss-* template).

Response

templatesarray

Array of template objects.

Template Object

namestring

Template identifier. Pass this exact value as model_name when creating a deployment.

typestring

Model type — "text2text" or "embedding".

metadataobject

Static metadata about the model.

Metadata Object

vendorstring

Upstream vendor / lab (e.g. "openai", "meta", "mistral").

context_window_kinteger

Context window in thousands of tokens.

size_bnumber

Parameter count in billions.

licensestring

SPDX license identifier (e.g. "apache-2.0", "llama-3").

flavorsarray

Deployable variants of the template.

Flavor Object

namestring

Flavor identifier. Pass as flavor_name when creating a deployment. Default is "base".

quantizationstring | null

Weight quantization (e.g. "fp8", "int4"), or null for the unquantized base flavor.

use_casesarray

Suggested use cases (e.g. ["chat", "code"]).

tagsarray

Free-form tags for filtering in a UI (e.g. ["long-context", "tool-use"]).

available_configurationsobject

Valid GPU configurations for this flavor.

Available Configurations

gpu_configurationsarray

List of { gpu_type, gpu_count, regions } entries. create-deployment requires the gpu_type + gpu_count + region triple to match one of these.

Request Example

curl "https://{organization}.blackbox.ai/dedicated/templates?type=text2text&model=gpt-oss" \
  -H "Authorization: Bearer $BLACKBOX_API_KEY"

import os
import requests

response = requests.get(
    "https://{organization}.blackbox.ai/dedicated/templates",
    headers={"Authorization": f"Bearer {os.environ['BLACKBOX_API_KEY']}"},
    params={"type": "text2text", "model": "gpt-oss"},
)

print(response.json())

const url = new URL("https://{organization}.blackbox.ai/dedicated/templates");
url.searchParams.set("type", "text2text");
url.searchParams.set("model", "gpt-oss");

const response = await fetch(url, {
  headers: { Authorization: `Bearer ${process.env.BLACKBOX_API_KEY}` },
});

const data = await response.json();
console.log(data);

Response Example

{
  "templates": [
    {
      "name": "openai/gpt-oss-20b",
      "type": "text2text",
      "metadata": {
        "vendor": "openai",
        "context_window_k": 128,
        "size_b": 20,
        "license": "apache-2.0"
      },
      "flavors": [
        {
          "name": "base",
          "quantization": null,
          "use_cases": ["chat", "code"],
          "tags": ["long-context", "tool-use"],
          "available_configurations": {
            "gpu_configurations": [
              {
                "gpu_type": "gpu-h100-sxm",
                "gpu_count": [1, 2, 4],
                "regions": ["eu-north1", "us-east1"]
              },
              {
                "gpu_type": "gpu-a100-80gb",
                "gpu_count": [1, 2],
                "regions": ["us-east1"]
              }
            ]
          }
        },
        {
          "name": "fp8",
          "quantization": "fp8",
          "use_cases": ["chat"],
          "tags": ["low-latency"],
          "available_configurations": {
            "gpu_configurations": [
              {
                "gpu_type": "gpu-h100-sxm",
                "gpu_count": [1],
                "regions": ["eu-north1", "us-east1"]
              }
            ]
          }
        }
      ]
    }
  ]
}