ExBifrost.Api.GenAIIntegration (ex_bifrost v0.1.0)

Copy Markdown View Source

API calls for all endpoints tagged GenAIIntegration.

Summary

Functions

gemini_cancel_batch(connection, batch_id, opts \\ [])

Cancel a batch job (Gemini format) Cancels a batch job. Gemini conventionally sends the request as /v1beta/batches/{batch_id}:cancel; the router matches the :cancel suffix into the batch_id path parameter.

gemini_count_tokens(connection, model, gemini_count_tokens_request, opts \\ [])

Count tokens (Gemini format) Counts tokens using Google Gemini API format.

gemini_create_cached_content(connection, gemini_create_cached_content_request, opts \\ [])

Create cached content (Gemini format) Creates a cached content entry that can be re-used across subsequent generate-content calls to reduce repeated prefix tokens.

gemini_delete_batch(connection, batch_id, opts \\ [])

Delete a batch job (Gemini format)

gemini_delete_cached_content(connection, cached_id, opts \\ [])

Delete cached content (Gemini format)

gemini_delete_file(connection, file_id, opts \\ [])

Delete file (Gemini format) Deletes a file in Google Gemini API format.

gemini_embed_content(connection, model, gemini_embedding_request, opts \\ [])

Embed content (Gemini format) Creates embeddings using Google Gemini API format.

gemini_generate_content(connection, model, gemini_generation_request, opts \\ [])

Generate content (Gemini format) Generates content using Google Gemini API format. The model is specified in the URL path.

gemini_generate_image(connection, model, gemini_generation_request, opts \\ [])

Generate image (Gemini format) For Imagen models, use the :predict suffix (e.g., imagen-3.0-generate-001:predict). For Gemini models, use :generateContent with generationConfig.responseModalities: ["IMAGE"] in the request body.

gemini_list_batches(connection, opts \\ [])

List batch jobs (Gemini format) Lists batch jobs in Gemini format. Supports pageSize / pageToken pagination.

gemini_list_cached_contents(connection, opts \\ [])

List cached content entries (Gemini format)

gemini_list_files(connection, opts \\ [])

List files (Gemini format) Lists uploaded files in Google Gemini API format.

gemini_list_models(connection, opts \\ [])

List models (Gemini format) Lists available models in Google Gemini API format.

gemini_retrieve_batch(connection, batch_id, opts \\ [])

Retrieve a batch job (Gemini format)

gemini_retrieve_cached_content(connection, cached_id, opts \\ [])

Retrieve cached content (Gemini format)

gemini_retrieve_file(connection, file_id, opts \\ [])

Retrieve file (Gemini format) Retrieves file metadata in Google Gemini API format. Note: This endpoint returns file metadata only. Direct file content download is not supported by Gemini Files API. Use the file.uri field from the response to access the file content.

gemini_retrieve_video_operation(connection, model, operation_id, opts \\ [])

Retrieve video generation operation (Gemini format) Polls the status of a long-running video generation operation produced by models/{model}:generateVideos. The Gemini SDK appends the operation name as a wildcard path segment. If the operation name contains /, it must be percent-encoded in the request path (for example, %2F) to remain conformant with OpenAPI 3.x path-parameter semantics.

gemini_stream_generate_content(connection, model, gemini_generation_request, opts \\ [])

Stream generate content (Gemini format) Streams content generation using Google Gemini API format. The model is specified in the URL path.

gemini_update_cached_content(connection, cached_id, gemini_update_cached_content_request, opts \\ [])

Update cached content (Gemini format) Updates the TTL or expiration time of a cached content entry. Only ttl or expireTime may be modified.

gemini_upload_file(connection, file, opts \\ [])

Upload file (Gemini format) Uploads a file using Google Gemini API format. This is a multipart upload with two parts: - "metadata": JSON object containing file metadata - "file": Binary file content Note: Direct file content download is not supported by Gemini Files API. Use the file.uri field from the response to access uploaded files.

vertex_rank(connection, vertex_rank_request, opts \\ [])

Rerank documents (Vertex Rank) Reranks records using Google Vertex AI's Ranking API. The request body follows the Vertex rankRecords schema.

Functions

gemini_cancel_batch(connection, batch_id, opts \\ [])

@spec gemini_cancel_batch(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Cancel a batch job (Gemini format) Cancels a batch job. Gemini conventionally sends the request as /v1beta/batches/{batch_id}:cancel; the router matches the :cancel suffix into the batch_id path parameter.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • batch_id (String.t): Batch identifier, optionally with a :cancel action suffix.
  • opts (keyword): Optional parameters

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_count_tokens(connection, model, gemini_count_tokens_request, opts \\ [])

@spec gemini_count_tokens(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiCountTokensRequest.t(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiCountTokens200Response.t()}
  | {:error, Tesla.Env.t()}

Count tokens (Gemini format) Counts tokens using Google Gemini API format.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Model name with action (e.g., gemini-pro:countTokens)
  • gemini_count_tokens_request (GeminiCountTokensRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiCountTokens200Response.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_create_cached_content(connection, gemini_create_cached_content_request, opts \\ [])

@spec gemini_create_cached_content(
  Tesla.Env.client(),
  ExBifrost.Model.GeminiCreateCachedContentRequest.t(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Create cached content (Gemini format) Creates a cached content entry that can be re-used across subsequent generate-content calls to reduce repeated prefix tokens.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • gemini_create_cached_content_request (GeminiCreateCachedContentRequest):
  • opts (keyword): Optional parameters
    • :"x-model-provider" (String.t): Optional provider override. Defaults to gemini. Use vertex to route to Vertex.

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_delete_batch(connection, batch_id, opts \\ [])

@spec gemini_delete_batch(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Delete a batch job (Gemini format)

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • batch_id (String.t):
  • opts (keyword): Optional parameters

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_delete_cached_content(connection, cached_id, opts \\ [])

@spec gemini_delete_cached_content(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Delete cached content (Gemini format)

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • cached_id (String.t):
  • opts (keyword): Optional parameters
    • :"x-model-provider" (String.t):

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_delete_file(connection, file_id, opts \\ [])

@spec gemini_delete_file(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, map()}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Delete file (Gemini format) Deletes a file in Google Gemini API format.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • file_id (String.t): File ID to delete
  • opts (keyword): Optional parameters

Returns

  • {:ok, map()} on success
  • {:error, Tesla.Env.t} on failure

gemini_embed_content(connection, model, gemini_embedding_request, opts \\ [])

@spec gemini_embed_content(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiEmbeddingRequest.t(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiEmbeddingResponse.t()}
  | {:error, Tesla.Env.t()}

Embed content (Gemini format) Creates embeddings using Google Gemini API format.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Model name with action (e.g., embedding-001:embedContent)
  • gemini_embedding_request (GeminiEmbeddingRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiEmbeddingResponse.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_generate_content(connection, model, gemini_generation_request, opts \\ [])

@spec gemini_generate_content(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiGenerationRequest.t(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiGenerationResponse.t()}
  | {:error, Tesla.Env.t()}

Generate content (Gemini format) Generates content using Google Gemini API format. The model is specified in the URL path.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Model name with action (e.g., gemini-pro:generateContent)
  • gemini_generation_request (GeminiGenerationRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiGenerationResponse.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_generate_image(connection, model, gemini_generation_request, opts \\ [])

@spec gemini_generate_image(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiGenerationRequest.t(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiGenerationResponse.t()}
  | {:error, Tesla.Env.t()}

Generate image (Gemini format) For Imagen models, use the :predict suffix (e.g., imagen-3.0-generate-001:predict). For Gemini models, use :generateContent with generationConfig.responseModalities: ["IMAGE"] in the request body.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Model name with action suffix. For Imagen models, use :predict (e.g., imagen-3.0-generate-001:predict). For Gemini models with image generation, use :generateContent (e.g., gemini-1.5-pro:generateContent).
  • gemini_generation_request (GeminiGenerationRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiGenerationResponse.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_list_batches(connection, opts \\ [])

@spec gemini_list_batches(
  Tesla.Env.client(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

List batch jobs (Gemini format) Lists batch jobs in Gemini format. Supports pageSize / pageToken pagination.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • opts (keyword): Optional parameters
    • :pageSize (integer()):
    • :pageToken (String.t):

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_list_cached_contents(connection, opts \\ [])

@spec gemini_list_cached_contents(
  Tesla.Env.client(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

List cached content entries (Gemini format)

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • opts (keyword): Optional parameters
    • :pageSize (integer()):
    • :pageToken (String.t):
    • :"x-model-provider" (String.t):

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_list_files(connection, opts \\ [])

@spec gemini_list_files(
  Tesla.Env.client(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiListFiles200Response.t()}
  | {:error, Tesla.Env.t()}

List files (Gemini format) Lists uploaded files in Google Gemini API format.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • opts (keyword): Optional parameters
    • :pageSize (integer()): Maximum number of files to return
    • :pageToken (String.t): Page token for pagination

Returns

  • {:ok, ExBifrost.Model.GeminiListFiles200Response.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_list_models(connection, opts \\ [])

@spec gemini_list_models(
  Tesla.Env.client(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiListModelsResponse.t()}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

List models (Gemini format) Lists available models in Google Gemini API format.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • opts (keyword): Optional parameters
    • :pageSize (integer()): Maximum number of models to return
    • :pageToken (String.t): Page token for pagination

Returns

  • {:ok, ExBifrost.Model.GeminiListModelsResponse.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_retrieve_batch(connection, batch_id, opts \\ [])

@spec gemini_retrieve_batch(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Retrieve a batch job (Gemini format)

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • batch_id (String.t):
  • opts (keyword): Optional parameters

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_retrieve_cached_content(connection, cached_id, opts \\ [])

@spec gemini_retrieve_cached_content(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Retrieve cached content (Gemini format)

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • cached_id (String.t):
  • opts (keyword): Optional parameters
    • :"x-model-provider" (String.t):

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_retrieve_file(connection, file_id, opts \\ [])

@spec gemini_retrieve_file(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, ExBifrost.Model.GeminiUploadFile200ResponseFile.t()}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Retrieve file (Gemini format) Retrieves file metadata in Google Gemini API format. Note: This endpoint returns file metadata only. Direct file content download is not supported by Gemini Files API. Use the file.uri field from the response to access the file content.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • file_id (String.t): File ID
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiUploadFile200ResponseFile.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_retrieve_video_operation(connection, model, operation_id, opts \\ [])

@spec gemini_retrieve_video_operation(
  Tesla.Env.client(),
  String.t(),
  String.t(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Retrieve video generation operation (Gemini format) Polls the status of a long-running video generation operation produced by models/{model}:generateVideos. The Gemini SDK appends the operation name as a wildcard path segment. If the operation name contains /, it must be percent-encoded in the request path (for example, %2F) to remain conformant with OpenAPI 3.x path-parameter semantics.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Video model identifier (e.g. veo-3.1-generate-preview).
  • operation_id (String.t): Operation name. If it contains /, it must be percent-encoded in the request path (for example, %2F).
  • opts (keyword): Optional parameters

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_stream_generate_content(connection, model, gemini_generation_request, opts \\ [])

@spec gemini_stream_generate_content(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiGenerationRequest.t(),
  keyword()
) ::
  {:ok, ExBifrost.Model.GeminiError.t()}
  | {:ok, ExBifrost.Model.GeminiGenerationResponse.t()}
  | {:error, Tesla.Env.t()}

Stream generate content (Gemini format) Streams content generation using Google Gemini API format. The model is specified in the URL path.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • model (String.t): Model name with action (e.g., gemini-pro:streamGenerateContent)
  • gemini_generation_request (GeminiGenerationRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, ExBifrost.Model.GeminiGenerationResponse.t} on success
  • {:error, Tesla.Env.t} on failure

gemini_update_cached_content(connection, cached_id, gemini_update_cached_content_request, opts \\ [])

@spec gemini_update_cached_content(
  Tesla.Env.client(),
  String.t(),
  ExBifrost.Model.GeminiUpdateCachedContentRequest.t(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Update cached content (Gemini format) Updates the TTL or expiration time of a cached content entry. Only ttl or expireTime may be modified.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • cached_id (String.t):
  • gemini_update_cached_content_request (GeminiUpdateCachedContentRequest):
  • opts (keyword): Optional parameters
    • :"x-model-provider" (String.t):

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure

gemini_upload_file(connection, file, opts \\ [])

@spec gemini_upload_file(Tesla.Env.client(), String.t(), keyword()) ::
  {:ok, ExBifrost.Model.GeminiUploadFile200Response.t()}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Upload file (Gemini format) Uploads a file using Google Gemini API format. This is a multipart upload with two parts: - "metadata": JSON object containing file metadata - "file": Binary file content Note: Direct file content download is not supported by Gemini Files API. Use the file.uri field from the response to access uploaded files.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • file (String.t):
  • opts (keyword): Optional parameters
    • :metadata (ExBifrost.Model.GeminiUploadFileRequestMetadata.t):

Returns

  • {:ok, ExBifrost.Model.GeminiUploadFile200Response.t} on success
  • {:error, Tesla.Env.t} on failure

vertex_rank(connection, vertex_rank_request, opts \\ [])

@spec vertex_rank(
  Tesla.Env.client(),
  ExBifrost.Model.VertexRankRequest.t(),
  keyword()
) ::
  {:ok, %{optional(String.t()) => any()}}
  | {:ok, ExBifrost.Model.GeminiError.t()}
  | {:error, Tesla.Env.t()}

Rerank documents (Vertex Rank) Reranks records using Google Vertex AI's Ranking API. The request body follows the Vertex rankRecords schema.

Parameters

  • connection (ExBifrost.Connection): Connection to server
  • vertex_rank_request (VertexRankRequest):
  • opts (keyword): Optional parameters

Returns

  • {:ok, %{}} on success
  • {:error, Tesla.Env.t} on failure