Create embeddings

Turns texts (or token IDs, or multimodal inputs for models that accept media) into embedding vectors, one per input in the order of input. Same format as OpenRouter and OpenAI; usage.cost is the EUR cost of the successful result attempt across its batches. Confirmed paid costs of earlier failed attempts remain charged and are additionally included in generation lookup and the durable billing receipt.

The gateway picks a European provider of the model (respecting provider preferences) and, if it fails, tries the next one. The body is validated completely before any provider is called; every problem is listed in error.metadata.issues. Unknown parameters are ignored and named in x-ignored-params. Responses are bounded like chat responses (32 MiB, see "Limits"): large batches of long vectors fit better with encoding_format: base64.

Administrative PII requirements are checked for each actual route, including fallbacks and retries. This endpoint cannot redact its input: a matching requirement returns invalid_parameter with a not_available issue at $.model before provider I/O.

POST https://api.tokenlane.de/v1/embeddings

Authorizations

[ [ { "name": "Authorization", "description": "API key as bearer token in the Authorization header: Authorization: Bearer sk-tl-v1-…. Existing sk-eu- keys remain valid. Missing, unknown or expired keys get 401 authentication.", "type": "http bearer", "location": "", "scopes": [] } ] ]

curl

curl --request POST \
  --url 'https://api.tokenlane.de/v1/embeddings' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "baai/bge-m3"
}'

node

import OpenAI from 'openai';

const client = new OpenAI({ baseURL: "https://api.tokenlane.de/v1", apiKey: process.env.TOKENLANE_API_KEY });
const response = await client.embeddings.create({
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "baai/bge-m3"
}, {});
console.log(response);

python

import os
import json
from openai import OpenAI

client = OpenAI(base_url="https://api.tokenlane.de/v1", api_key=os.environ["TOKENLANE_API_KEY"])
body = json.loads(r'''{
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "baai/bge-m3"
}''')
response = client.embeddings.create(
    model=body["model"],
    input=body["input"],
    encoding_format=body["encoding_format"]
)
print(response)

vercel

// This operation uses HTTP directly.
const response = await fetch("https://api.tokenlane.de/v1/embeddings", {
  method: "POST",
  headers: {
    "Authorization": "Bearer " + process.env.TOKENLANE_API_KEY,
    "Content-Type": "application/json",
  },
  body: `{
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "baai/bge-m3"
}`,
});
console.log(await response.text());

header: x-session-id (optional)

Groups the generation into a session, immediately available from GET /v1/sessions/{id}. Use 1 to 256 ASCII letters, digits, '-', '_', '.' or ':'; an ID cannot consist only of dots. A non-null session_id in the body takes precedence. Never sent to the provider.

s0: string | null

{}

header: Idempotency-Key (optional)

Explicit opt-in to encrypted response replay across gateway nodes. One 1..=255 printable ASCII value per logical operation, scoped to this API key ID. Reuse the same method, path and normalized body. Completed responses are retained for at most 10 minutes; replay is free and creates no generation. Conflicting content or unavailable replay returns 409. A successful replay carries Idempotent-Replayed: true and preserves the original request and generation IDs. See /docs/guides/idempotency.

Body: application/json

{
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "baai/bge-m3"
}

s14: object

Texts (or token IDs, or multimodal inputs) to turn into embedding vectors, in OpenRouter's format; OpenAI and OpenRouter clients work by changing the base URL and the API key.

The body is validated before any provider is called, and every problem is listed in error.metadata.issues. Limits: at most 2048 inputs, each within the model's context length (estimated at 4 bytes of text per token, 400 context_length_exceeded above it), and the 50 MiB body with its structure limit (see "Limits" in the introduction). Unknown parameters are ignored and named in x-ignored-params.

{}

{
  "encoding_format": "float",
  "input": [
    "Lisbon is the capital of Portugal.",
    "Porto lies on the Douro."
  ],
  "model": "mistralai/mistral-embed"
}
  • dimensions (optional): s291

s291: integer

Length of the vectors, for models that can shorten them.

{"minimum":"1"}

1024
  • encoding_format (optional): s292

s292: string

float (default) returns arrays of numbers, base64 the little-endian 32-bit floats as base64 strings, about a quarter of the size; use it for large batches.

"float", "base64"

{}

"float"
  • input (required): s293

s293: anyOf

Text, token, or multimodal input(s) to embed

{}

"Lisbon is the capital of Portugal."

anyOf: s1057

s1057: string

{"minLength":"1"}

anyOf: s1058

s1058: array

{}

anyOf: s1059

s1059: array

{}

s1251: number

{}

anyOf: s1060

s1060: array

{}

s1252: array

{}

anyOf: s1061

s1061: array

{}

s1253: object

One multimodal input, for models that accept its media (architecture.input_modalities in GET /v1/models).

{}

{
  "content": [
    {
      "text": "A cat",
      "type": "text"
    },
    {
      "image_url": {
        "url": "https://upload.wikimedia.org/wikipedia/commons/3/3a/Cat03.jpg"
      },
      "type": "image_url"
    }
  ]
}
  • content (required): s1335

s1335: array

{"minItems":"1"}

s1349: oneOf

{}

oneOf: s1351

s1351: object

{}

  • text (required): s1057
  • type (required): s1089

s1089: string

"text"

{}

oneOf: s1352

s1352: object

{}

  • image_url (required): s1353

s1353: object

{}

  • url (required): s7

s7: string

{}

  • type (required): s1229

s1229: string

"image_url"

{}

oneOf: s94

s94: object

Audio of a multimodal embeddings input.

{}

{
  "input_audio": {
    "data": "data:audio/wav;base64,UklGRiQAAABXQVZFZm10IBAAAAA=",
    "format": "wav"
  },
  "type": "input_audio"
}
  • input_audio (required): s566

s566: object

Audio of a multimodal embeddings input.

{}

{
  "data": "data:audio/wav;base64,UklGRiQAAABXQVZFZm10IBAAAAA=",
  "format": "wav"
}
  • data (required): s907

s907: string

The media as a base64 data URL (data:<media type>;base64,<data>) or as plain base64.

{}

  • format (optional): s908

s908: string

Format of the media, for example wav, mp4 or pdf.

{}

"wav"
  • type (required): s567

s567: string

Kind of the part; names the field that carries the media.

"input_audio"

{}

"input_audio"

oneOf: s96

s96: object

A video of a multimodal embeddings input.

{}

{
  "input_video": {
    "data": "data:video/mp4;base64,AAAAGGZ0eXBtcDQyAAAAAGlzb20=",
    "format": "mp4"
  },
  "type": "input_video"
}
  • input_video (required): s570

s570: object

A video of a multimodal embeddings input.

{}

{
  "data": "data:audio/wav;base64,UklGRiQAAABXQVZFZm10IBAAAAA=",
  "format": "wav"
}
  • data (required): s907

The media as a base64 data URL (data:<media type>;base64,<data>) or as plain base64.

  • format (optional): s908

Format of the media, for example wav, mp4 or pdf.

  • type (required): s571

s571: string

Kind of the part; names the field that carries the media.

"input_video"

{}

"input_video"

oneOf: s95

s95: object

A document of a multimodal embeddings input.

{}

{
  "input_file": {
    "data": "data:application/pdf;base64,JVBERi0xLjQK",
    "format": "pdf"
  },
  "type": "input_file"
}
  • input_file (required): s568

s568: object

A document of a multimodal embeddings input.

{}

{
  "data": "data:audio/wav;base64,UklGRiQAAABXQVZFZm10IBAAAAA=",
  "format": "wav"
}
  • data (required): s907

The media as a base64 data URL (data:<media type>;base64,<data>) or as plain base64.

  • format (optional): s908

Format of the media, for example wav, mp4 or pdf.

  • type (required): s569

s569: string

Kind of the part; names the field that carries the media.

"input_file"

{}

"input_file"
  • input_type (optional): s294

s294: string

How the input will be used, for models that embed queries and documents differently, for example search_query or search_document. Passed to the provider as sent.

{}

"search_document"
  • metadata (optional): s295

s295: any

Local metadata: at most 16 string pairs, keys 1..64 and values 0..512 characters.

{}

  • model (required): s296

s296: string

Embedding model, with the same ID as on OpenRouter, or an alias of the gateway. Chat models cannot embed: their ID fails with 404 no_endpoint. An unknown ID fails with 400 model_not_found and a suggestion in remedy_hint.

{}

"mistralai/mistral-embed"
  • provider (optional): s297

s297: object

Routing preferences, as for chat completions: only, ignore, order, allow_fallbacks, require_parameters, sort and max_price (EUR per million tokens). If a provider fails, the next one is tried.

{}

  • session_id (optional): s249

s249: string | null

Groups related requests in generation records and GET /v1/sessions/{id}. Use 1 to 256 ASCII letters, digits, -, _, . or :; not only dots. A non-null body value takes precedence over x-session-id. Never sent to the provider.

{"minLength":"1","maxLength":"256"}

"session-1234"
  • trace (optional): s252

s252: object

Trace metadata for observability; never sent to the provider.

{}

  • user (optional): s298

s298: string

An identifier of the end user, retained locally; sent only as a tenant-scoped hash when the route explicitly supports user.

{}

"user-1234"

Response 200: application/json

One vector per input, with token counts and the cost in EUR in usage. Always JSON: embeddings are never streamed (text/event-stream is listed only for compatibility with OpenRouter's description).

{
  "data": [
    {
      "embedding": [
        0.0023064255,
        -0.009327292,
        0.015797347
      ],
      "index": 0,
      "object": "embedding"
    }
  ],
  "id": "gen-01k7tm9qz3f5c8x2v6b4n0m7hd",
  "model": "mistralai/mistral-embed",
  "object": "list",
  "provider": "mistral",
  "usage": {
    "cost": 8e-7,
    "cost_details": {
      "currency": "EUR",
      "upstream_inference_completions_cost": 0,
      "upstream_inference_cost": 8e-7,
      "upstream_inference_prompt_cost": 8e-7
    },
    "currency": "EUR",
    "is_byok": false,
    "prompt_tokens": 8,
    "total_tokens": 8
  }
}

s15: object

The embedding vectors, in OpenRouter's format, with the cost in EUR.

{}

{
  "data": [
    {
      "embedding": [
        0.0023064255,
        -0.009327292,
        0.015797347
      ],
      "index": 0,
      "object": "embedding"
    }
  ],
  "id": "gen-01k7tm9qz3f5c8x2v6b4n0m7hd",
  "model": "mistralai/mistral-embed",
  "object": "list",
  "provider": "mistral",
  "usage": {
    "cost": 8e-7,
    "cost_details": {
      "currency": "EUR",
      "upstream_inference_completions_cost": 0,
      "upstream_inference_cost": 8e-7,
      "upstream_inference_prompt_cost": 8e-7
    },
    "currency": "EUR",
    "is_byok": false,
    "prompt_tokens": 8,
    "total_tokens": 8
  }
}
  • data (required): s299

s299: array

One vector per input, in the order of the inputs (index).

{}

s117: object

The vector of one input.

{}

{
  "embedding": [
    0.0023064255,
    -0.009327292,
    0.015797347
  ],
  "index": 0,
  "object": "embedding"
}
  • embedding (required): s624

s624: anyOf

The vector: numbers, or a base64 string with encoding_format: base64

{}

[
  0.0023064255,
  -0.009327292,
  0.015797347
]

anyOf: s1157

s1157: array

{}

anyOf: s7

  • index (required): s625

s625: integer

Position of the input in input.

int32

{"minimum":"0"}

0
  • object (required): s626

s626: string

Always embedding.

"embedding"

{}

"embedding"
  • id (required): s300

s300: string

ID of this generation; GET /v1/generation?id= shows its tokens, cost and timing.

{}

"gen-01k7tm9qz3f5c8x2v6b4n0m7hd"
  • model (required): s301

s301: string

The model that embedded, as a public model ID.

{}

"mistralai/mistral-embed"
  • object (required): s302

s302: string

Always list.

"list"

{}

"list"
  • usage (optional): s303

s303: object

Token counts and EUR cost of the successful result attempt, including all its batches, if the provider reported input tokens. Earlier paid attempts that failed before a successful fallback remain charged and are included in GET /v1/generation and the durable billing receipt, not this result's usage. Without confirmed input tokens, cost is missing rather than zero.

{}

{
  "cost": 8e-7,
  "currency": "EUR",
  "is_byok": false,
  "prompt_tokens": 8,
  "total_tokens": 8
}
  • cost (optional): s627

s627: number

EUR cost of the successful result attempt across all its batches. Generation lookup and the durable receipt additionally include confirmed costs of earlier failed attempts.

double

{}

8e-7
  • currency (optional): s628

s628: string

Always EUR.

{}

"EUR"
  • is_byok (optional): s629

s629: boolean

Whether the client's own provider key was used. Always false: the gateway has no BYOK.

{}

  • prompt_tokens (required): s630

s630: integer

Input tokens, as the provider counted them.

int64

{"minimum":"0"}

8
  • total_tokens (required): s631

s631: integer

All tokens; the same as prompt_tokens.

int64

{"minimum":"0"}

8
  • * (optional): s632

s632: any

Public cost_details split input/output costs in EUR, with original-currency and FX details when applicable. Unknown provider usage fields are removed.

{}

  • * (optional): s304

s304: any

Public extension provider; unknown provider envelope fields are removed.

{}

Response 200: text/event-stream

One vector per input, with token counts and the cost in EUR in usage. Always JSON: embeddings are never streamed (text/event-stream is listed only for compatibility with OpenRouter's description).

YOUR_VALUE

Response 400: application/json

Recognized upstream 400/422 request rejections return non-retryable client errors with safe public field paths. Invalid request: broken JSON (invalid_request), invalid fields such as an empty or missing input or more than 2048 inputs, with every problem in issues (invalid_parameter), an unknown model ID with a suggestion in remedy_hint (model_not_found), or an input too long for the model (context_length_exceeded)

{
  "error": {
    "code": 400,
    "message": "deepseek/deepseek-v4.1-flsh is not a valid model ID",
    "metadata": {
      "docs_url": "/docs#description/model-not-found",
      "issues": [
        {
          "code": "unknown_model",
          "message": "deepseek/deepseek-v4.1-flsh is not a valid model ID",
          "path": "$.model"
        }
      ],
      "remedy_hint": "Did you mean deepseek/deepseek-v4.1-flash?",
      "request_id": "req_01k7tm9qz3f5c8x2v6b4n0m7hd",
      "retryable": false,
      "type": "model_not_found"
    }
  },
  "user_id": "org_01k7tm2c4e6g8j0k2m4p6r8t0v"
}

s3: object

Every response with an HTTP status of 400 or higher has this shape, OpenRouter's error format. Everything the gateway adds is inside error.metadata, so OpenRouter clients see nothing unfamiliar. See "Errors" in the introduction for all error types.

{}

{
  "error": {
    "code": 400,
    "message": "deepseek/deepseek-v4.1-flsh is not a valid model ID",
    "metadata": {
      "docs_url": "/docs#description/model-not-found",
      "issues": [
        {
          "code": "unknown_model",
          "message": "deepseek/deepseek-v4.1-flsh is not a valid model ID",
          "path": "$.model"
        }
      ],
      "remedy_hint": "Did you mean deepseek/deepseek-v4.1-flash?",
      "request_id": "req_01k7tm9qz3f5c8x2v6b4n0m7hd",
      "retryable": false,
      "type": "model_not_found"
    }
  },
  "user_id": "org_01k7tm2c4e6g8j0k2m4p6r8t0v"
}
  • error (required): s238

s238: object

What went wrong.

{}

{
  "code": 400,
  "message": "temperature must be at most 2, got 5 (and 1 more)",
  "metadata": {
    "docs_url": "/docs#description/invalid-parameter",
    "issues": [
      {
        "code": "out_of_range",
        "message": "temperature must be at most 2, got 5",
        "path": "$.temperature"
      },
      {
        "code": "invalid_value",
        "message": "messages[1].role must be one of system, developer, user, assistant, tool, got \"bot\"",
        "path": "$.messages[1].role"
      }
    ],
    "request_id": "req_01k7tm9qz3f5c8x2v6b4n0m7hd",
    "retryable": false,
    "type": "invalid_parameter"
  }
}
  • code (required): s633

s633: integer

The HTTP status of the response, as a number.

int32

{"minimum":"0"}

400
  • message (required): s634

s634: string

What went wrong, in English. With several problems: the first one and how many more follow; all of them are in metadata.issues. The text may change; match on metadata.type instead.

{}

"temperature must be at most 2, got 5"
  • metadata (optional): s635

s635: object

Details for programs: error type, request ID, every problem, a suggested fix and, for provider errors, which provider failed. Always present on gateway errors.

{}

{
  "docs_url": "/docs#description/rate-limited",
  "provider_error": {
    "category": "rate_limited",
    "provider_name": "deepinfra",
    "retry_after": 2,
    "retryable": true,
    "status": 429
  },
  "provider_name": "deepinfra",
  "remedy_hint": "Retry after the Retry-After interval, or allow fallbacks to other providers (provider.allow_fallbacks).",
  "request_id": "req_01k7tm9qz3f5c8x2v6b4n0m7hd",
  "retry_after": 2,
  "retryable": true,
  "type": "rate_limited"
}
  • available_providers (optional): s639

s639: array

Providers that serve the model (when provider.only excluded all of them).

{}

[
  "deepinfra",
  "open-inference"
]
  • docs_url (required): s640

s640: string

Link to the section of this error type in the documentation.

{}

"/docs#description/model-not-found"
  • failed_routing_step (optional): s641

s641: string

The routing step that left no route (routing errors only).

{}

"Filter by Max Price"
  • issues (optional): s642

s642: array

Every problem found in the request, so all can be fixed at once (at most 20); for routing errors one entry per excluded route, for no_provider_available one per failed attempt.

{}

s123: object

One problem in the request, one reason a route was excluded, or one failed attempt.

{}

{
  "code": "invalid_url",
  "message": "messages[2].content[0].image_url.url must be an http(s) URL or a base64 data URL",
  "path": "$.messages[2].content[0].image_url.url"
}
  • code (required): s636

s636: string

Kind of problem, stable for programs.

"required", "invalid_type", "invalid_value", "out_of_range", "empty", "invalid_url", "invalid_base64", "unsupported_media_type", "media_type_mismatch", "media_too_large", "too_many_media", "unsupported_media_transport", "media_fetch_failed", "unknown_tool", "unknown_model", "not_available", "not_allowed", "route_excluded", "bridge_failed", "provider_failed"

{}

"out_of_range"
  • message (required): s637

s637: string

The problem in English, naming the field and, if short, the value sent.

{}

"temperature must be at most 2, got 5"
  • path (required): s638

s638: string

Where the problem is: JSONPath of the field in the request body, for example $.messages[2].content[0].image_url.url, or query.<name> for a query parameter.

{}

"$.temperature"
  • provider_error (optional): s643

s643: oneOf

Safe provider diagnostics, identical for HTTP errors and terminal stream events.

{}

oneOf: s1158

s1158: object

Provider diagnostics without provider-authored text. HTTP and terminal SSE errors use this same structure; arbitrary upstream messages and metadata are discarded.

{}

{
  "category": "rate_limited",
  "provider_name": "deepinfra",
  "retry_after": 2,
  "retryable": true,
  "status": 429
}
  • category (required): s1299

s1299: string

Gateway classification from HTTP status and recognized dialect identifiers.

"rate_limited", "overloaded", "invalid_request", "context_length", "content_filtered", "authentication", "billing", "timeout", "unavailable", "unknown"

{}

"rate_limited"
  • code (optional): s1300

s1300: string

Exact identifier from the configured dialect's fixed list; arbitrary values vanish.

{}

"rate_limit_exceeded"
  • provider_name (required): s1301

s1301: string

Configured provider name, never a name supplied by the upstream body.

{}

"deepinfra"
  • retry_after (optional): s1302

s1302: integer

Bounded seconds from the upstream Retry-After header, when present and numeric.

int64

{"minimum":"0"}

2
  • retryable (required): s1303

s1303: boolean

Whether the classified failure can be transient. Unknown failures are not promised retryable.

{}

  • status (optional): s1304

s1304: integer

Actual HTTP status received from the upstream, including 200 for native errors. Absent when no HTTP status is available.

int32

{"minimum":"0"}

429

oneOf: s588

s588: null

{}

  • provider_name (optional): s644

s644: string

The provider that failed, as in the response field provider (provider errors only).

{}

"deepinfra"
  • raw (optional): s645

s645: string

Reserved OpenRouter compatibility field. The gateway never fills it: provider free text is deliberately withheld. Use provider_error instead.

{}

""
  • remedy_hint (optional): s646

s646: string

A concrete fix, for example the closest model ID to a mistyped one.

{}

"Did you mean deepseek/deepseek-v4.1-flash?"
  • request_id (required): s647

s647: string

ID of the request, the same as the x-request-id response header. Quote it when contacting support.

{}

"req_01k7tm9qz3f5c8x2v6b4n0m7hd"
  • requested_providers (optional): s648

s648: array

Providers named in provider.only (when it excluded all routes).

{}

[
  "scaleway"
]
  • retry_after (optional): s649

s649: integer

Seconds to wait before retrying; also sent as the Retry-After header (429 and 503).

int64

{"minimum":"0"}

2
  • retryable (required): s650

s650: boolean

Whether sending the identical request again can succeed (for example after a timeout or rate limit), as opposed to errors that need a changed request.

{}

  • routing_funnel (optional): s651

s651: array

How routing narrowed down the providers: each step with the number of routes left (routing errors only, same format as OpenRouter).

{}

s200: object

One routing step, as in OpenRouter's routing_funnel.

{}

{
  "endpoint_count": 0,
  "reason": "provider.max_price",
  "step": "Filter by Max Price"
}
  • endpoint_count (required): s979

s979: integer

Number of routes (providers of the model) left after the step.

{"minimum":"0"}

2
  • reason (optional): s980

s980: string

The request field the step filtered on.

{}

"provider.only"
  • step (required): s981

s981: string

Name of the step, as at OpenRouter.

{}

"Initial Endpoints"
  • type (required): s652

s652: string

Stable error type to match on in code; see "Errors" in the introduction. Messages may change, the type does not.

"invalid_request", "invalid_parameter", "unsupported_parameter", "context_length_exceeded", "authentication", "insufficient_credits", "workspace_budget_exceeded", "permission_denied", "moderation", "model_not_found", "no_endpoint", "not_found", "method_not_allowed", "timeout", "payload_too_large", "media_fetch_failed", "rate_limited", "provider_error", "idempotency_conflict", "idempotency_in_progress", "idempotency_unavailable", "provider_invalid_response", "no_provider_available", "byok_invalid_credential", "byok_rate_limited", "byok_unavailable", "feature_unavailable", "internal"

{}

"invalid_parameter"
  • user_id (optional): s239

s239: string | null

Organization of the API key, if the key was recognized.

{}

"org_01k7tm2c4e6g8j0k2m4p6r8t0v"

Response 401: application/json

Missing, invalid or expired API key (authentication)

Response 402: application/json

The API key spending limit, the workspace budget (workspace_budget_exceeded) or the organization prepaid balance cannot cover the estimated request reservation, including fees, media and concurrent reservations (insufficient_credits for key and organization). No paid upstream call starts without a reservation.

Response 403: application/json

The organization, workspace or key policy does not allow the model (permission_denied); remedy_hint names the models it may use

Response 404: application/json

The model exists, but none of its providers fits the request: it is not an embedding model, or provider filters, max_price or media exclude every provider. issues names the reason per provider (no_endpoint)

Response 408: application/json

The provider did not connect or did not finish its response within the time limit, or the request body stopped arriving (timeout)

Response 409: application/json

idempotency_conflict: this key has different request content; use a new key. idempotency_in_progress: the original is running; respect Retry-After. idempotency_unavailable: no safe replay is possible; do not silently recompute. See remedy_hint.

Response 413: application/json

The request body is larger than 50 MiB, or its parsed JSON structure excluding JSON string values would need more than 20 MiB (payload_too_large)

Response 429: application/json

Customer quota or upstream rate limit exhausted (rate_limited)

Response 500: application/json

Unexpected gateway error (internal)

Response 501: application/json

feature_unavailable: A required feature is permanently unconfigured on this gateway, for example encrypted idempotency replay without its encryption key, PII redaction without a detector, or BYOK import without import keys. Ask the operator to configure it or remove the optional feature from the request. Retrying the identical request cannot enable the feature: retryable is false and no Retry-After header is sent. Temporary cache, database, detector or provider outages, pending imports and missing or revoked individual customer credentials retain their existing error types.

Response 502: application/json

The only provider tried failed (provider_error, with provider_name and structured provider_error diagnostics; provider text is withheld) or sent a response that is not a list of embeddings, larger than 32 MiB or too complex to parse (provider_invalid_response)

Response 503: application/json

Every provider failed, the gateway is busy with other large request bodies, or this gateway instance is shutting down (no_provider_available)