POST/api/orgs/{org_id}/provider/requests

File a change request

add_models applies at once when two gates pass: the manifest review against the live catalog, then a real streamed completion against your endpoint for every wire id (with the stored house key, or smoke_api_key, which authenticates the probe only and is never stored). Either failure answers 400 and lists nothing; new lanes land unlisted, so list each one afterwards. update_prices and update_model also apply at once (update_model at most once per 24 hours, else 429); update_limits lands pending for platform review. Org admin session or the org's own key.

Authentication: Bearer API key in the Authorization header

Path parameters

  • org_idstringrequired

    Your organization id (uuid). GET /api/whoami returns it.

Request body

  • kindenumrequired
    "add_models""update_prices""update_limits""update_model"
  • payloadobjectrequired

    The shape kind selects.

    One of

    add_models

    • modelsarray<any>required

      Wire ids must be unique, and no two entries may propose the same new model slug.

      models
      • wire_idstringrequired

        The model id your endpoint accepts in model.

      • bindingobjectrequired

        Bind to a catalog model (existing) or propose a new one (new).

        One of

        object

        • kindstringrequired
        • slugstringrequired

          An existing catalog model slug.

        object

        • kindstringrequired
        • modelobjectrequired
          model
          • slugstringrequired
          • display_namestringrequired
          • descriptionstring | null
          • context_windowinteger | null
          • max_output_tokensinteger | null
          • input_modalitiesarray<enum>
            Default: ["text"]
          • huggingface_urlstring | null
          • release_urlstring | null

            https URL with a real dotted host.

      • pricesobjectrequired

        Per-million-token rates in integer nano-USD (1 USD = 1,000,000,000 nano-USD). Each rate is at most 1,000,000,000,000 ($1,000 per million tokens).

        prices
        • input_nano_usd_per_millionintegerrequired
        • output_nano_usd_per_millionintegerrequired
        • cached_input_nano_usd_per_millioninteger | null

          Cached input rate; null means no cached rate.

        • reasoning_nano_usd_per_millioninteger | null
        • effective_atstring | null

          Optional timestamp with a timezone offset.

      • capabilitiesobject

        Declared capability booleans. Keys must come from the gateway's capability vocabulary (for example supports_streaming, supports_parallel_tool_calls, supports_structured_text, supports_image_input, reports_cached_input_tokens, reports_reasoning_tokens); unknown keys are refused. A lane needs supports_streaming: true to be listed.

        Default: {}
      • context_windowinteger | null
      • regionstring | null
      • deprecationobject | null
        deprecation
        • eol_atstringrequired
        • replacement_slugstring | null
      • data_policyobject | null

        Per-model data handling. Required on every model in an application.

        One of

        object

        • retains_promptsbooleanrequired
        • retention_daysinteger | null

          Days prompts are kept.

        • trains_on_databooleanrequired
        • zero_data_retentionbooleanrequired
        • policy_urlstring | null

          Public https URL of the policy.

      • dispatchobject | null
        One of

        object

        • scale_up_secondsinteger | null

          Warm-up time before saturation counts against the lane.

        • overflow_policyenum

          What happens once saturated.

          "reject""throttle"null
        • max_concurrencyinteger | null

          Declared cap on in-flight requests for this model.

      • rpminteger | null

        Requests per minute dispatch limit for the new lane; omit for the default.

      • tpminteger | null

        Tokens per minute dispatch limit for the new lane; omit for the default.

      • service_tier_pricing_enabledboolean | null
      • service_tier_ratesobject | null
        One of

        object

        • flexobject | null
          One of

          object

          • inputinteger | null
          • cached_inputinteger | null
          • cache_creation_inputinteger | null
          • cache_creation_1h_inputinteger | null
          • outputinteger | null
          • reasoninginteger | null
          • long_contextobject | null
            long_context
            • input_threshold_tokensintegerrequired
            • inputinteger | null
            • cached_inputinteger | null
            • cache_creation_inputinteger | null
            • cache_creation_1h_inputinteger | null
            • outputinteger | null
            • reasoninginteger | null
        • priorityobject | null
          One of

          object

          • inputinteger | null
          • cached_inputinteger | null
          • cache_creation_inputinteger | null
          • cache_creation_1h_inputinteger | null
          • outputinteger | null
          • reasoninginteger | null
          • long_contextobject | null
            long_context
            • input_threshold_tokensintegerrequired
            • inputinteger | null
            • cached_inputinteger | null
            • cache_creation_inputinteger | null
            • cache_creation_1h_inputinteger | null
            • outputinteger | null
            • reasoninginteger | null

    update_prices

    • changesarray<object>required
      changes
      • wire_idstringrequired

        A wire id one of your live lanes serves.

      • pricesobjectrequired

        Per-million-token rates in integer nano-USD (1 USD = 1,000,000,000 nano-USD). Each rate is at most 1,000,000,000,000 ($1,000 per million tokens).

        prices
        • input_nano_usd_per_millionintegerrequired
        • output_nano_usd_per_millionintegerrequired
        • cached_input_nano_usd_per_millioninteger | null

          Cached input rate; null means no cached rate.

        • reasoning_nano_usd_per_millioninteger | null
        • effective_atstring | null

          Optional timestamp with a timezone offset.

      • service_tier_pricing_enabledboolean | null

        true requires service_tier_rates, false clears tier pricing, null leaves it unchanged.

      • service_tier_ratesobject | null
        One of

        object

        • flexobject | null
          One of

          object

          • inputinteger | null
          • cached_inputinteger | null
          • cache_creation_inputinteger | null
          • cache_creation_1h_inputinteger | null
          • outputinteger | null
          • reasoninginteger | null
          • long_contextobject | null
            long_context
            • input_threshold_tokensintegerrequired
            • inputinteger | null
            • cached_inputinteger | null
            • cache_creation_inputinteger | null
            • cache_creation_1h_inputinteger | null
            • outputinteger | null
            • reasoninginteger | null
        • priorityobject | null
          One of

          object

          • inputinteger | null
          • cached_inputinteger | null
          • cache_creation_inputinteger | null
          • cache_creation_1h_inputinteger | null
          • outputinteger | null
          • reasoninginteger | null
          • long_contextobject | null
            long_context
            • input_threshold_tokensintegerrequired
            • inputinteger | null
            • cached_inputinteger | null
            • cache_creation_inputinteger | null
            • cache_creation_1h_inputinteger | null
            • outputinteger | null
            • reasoninginteger | null

    update_limits

    • max_concurrencyinteger | null
    • advertised_tpsnumber | null

      Advertised output tokens per second.

    update_model

    • wire_idstringrequired
    • pricesobject | null
      One of

      object

      • input_nano_usd_per_millionintegerrequired
      • output_nano_usd_per_millionintegerrequired
      • cached_input_nano_usd_per_millioninteger | null

        Cached input rate; null means no cached rate.

      • reasoning_nano_usd_per_millioninteger | null
      • effective_atstring | null

        Optional timestamp with a timezone offset.

    • data_policyobject | null
      One of

      object

      • retains_promptsbooleanrequired
      • retention_daysinteger | null

        Days prompts are kept.

      • trains_on_databooleanrequired
      • zero_data_retentionbooleanrequired
      • policy_urlstring | null

        Public https URL of the policy.

    • dispatchobject | null
      One of

      object

      • scale_up_secondsinteger | null

        Warm-up time before saturation counts against the lane.

      • overflow_policyenum

        What happens once saturated.

        "reject""throttle"null
      • max_concurrencyinteger | null

        Declared cap on in-flight requests for this model.

    • requests_per_minuteinteger | null

      Declared per-account requests per minute (observed, not enforced).

    • tokens_per_minuteinteger | null

      Declared per-account tokens per minute (observed, not enforced).

  • effective_atstring | null

    Recorded on a request that lands pending (update_limits). Must carry a timezone offset.

  • smoke_api_keystring | null

    Endpoint key for the add_models smoke when no house key is stored. Used for the probe only, never stored. Ignored for other kinds.

Responses

  • 201

    The recorded request: status: applied for add_models, update_prices and update_model, status: pending for update_limits.

    Response
    • idstringrequired
    • kindenumrequired
      "add_models""update_prices""update_limits""update_model"
    • statusenumrequired
      "pending""approved""rejected""applied""failed"
    • payloadobjectrequired

      The validated payload as stored.

    • effective_atstring | nullrequired
    • review_notestring | nullrequired

      The reviewer's note on a reviewed request.

    • errorstring | nullrequired

      Why an approved request failed to apply.

    • created_atstringrequired
    • applied_atstring | nullrequired
  • 400

    Refused: the manifest review found errors (request refused: <code>: <message>; ..., for example unknown-wire-id on a price change), the smoke test failed (code: smoke_failed, the upstream error per wire id verbatim), or the provider has no public https base_url to probe.

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

  • 401

    Missing, unknown or revoked API key.

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

  • 403

    The key is valid but may not perform this action (for example an inference key on a provisioning-key route, or a feature outside your plan).

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

  • 404

    The resource does not exist or belongs to another organization.

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

  • 409

    An add_models smoke needs a key and none is available (a house_key provider with no stored key and no smoke_api_key), or the catalog write conflicted.

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

  • 422

    The body failed validation, payload does not match kind, an add_models batch exceeds 200 models, or the catalog refused the write.

  • 429

    An update_model edit was filed less than 24 hours after the previous one.

    Response
    • errorstringrequired

      Human-readable message.

    • codestring

      Stable machine reason, when the route defines one (for example pro_required).

    • actionstring

      Stable remediation action, when the route defines one.

    • detailsobject

      Customer-safe string facts the caller needs to act (for example a suggested identifier).

Request

curl -X POST https://api.experientiallabs.ai/api/orgs/3f6c2a1e-8b7d-4c2a-9e51-6a0d4b2f7c10/provider/requests \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "kind": "add_models",
  "payload": {
    "models": [
      {
        "wire_id": "meta-llama/llama-4.1-70b-instruct",
        "binding": {
          "kind": "existing",
          "slug": "llama-4.1-70b-instruct"
        },
        "prices": {
          "input_nano_usd_per_million": 350000000,
          "output_nano_usd_per_million": 900000000,
          "cached_input_nano_usd_per_million": 35000000
        },
        "capabilities": {
          "supports_streaming": true,
          "supports_parallel_tool_calls": true,
          "reports_cached_input_tokens": true
        },
        "context_window": 131072,
        "data_policy": {
          "retains_prompts": false,
          "trains_on_data": false,
          "zero_data_retention": true
        },
        "rpm": 600,
        "tpm": 2000000
      }
    ]
  },
  "smoke_api_key": "sk-acme-probe-7d2c9e4b1a6f"
}'

Response

{
  "id": "8a4c2e61-7b9d-4f03-a1e5-c6d8b0f2e394",
  "kind": "add_models",
  "status": "applied",
  "payload": {
    "models": [
      {
        "wire_id": "meta-llama/llama-4.1-70b-instruct",
        "binding": {
          "kind": "existing",
          "slug": "llama-4.1-70b-instruct"
        },
        "prices": {
          "input_nano_usd_per_million": 350000000,
          "output_nano_usd_per_million": 900000000,
          "cached_input_nano_usd_per_million": 35000000,
          "reasoning_nano_usd_per_million": null,
          "effective_at": null
        }
      }
    ]
  },
  "effective_at": null,
  "review_note": null,
  "error": null,
  "created_at": "2026-11-19T10:12:44.903Z",
  "applied_at": "2026-11-19T10:12:44.903Z"
}

https://api.experientiallabs.ai/api/orgs/3f6c2a1e-8b7d-4c2a-9e51-6a0d4b2f7c10/provider/requests