Skip to main content
BytePlus Ark logo
  • Category: ai models
  • Auth: API_KEY
  • Composio-managed OAuth available? N/A
  • Tools: 17
  • Triggers: 0
  • Slug: BYTE_PLUS_ARK
  • Version: 20261101_00

Tools

Slug: BYTE_PLUS_ARK_CANCEL_OR_DELETE_VIDEO_TASKIrreversibly cancel a video generation task when it is queued, or permanently delete its task record when it is terminal. BytePlus Ark chooses the effect from the task’s current status; the caller cannot select cancel versus delete. Check the task with Get Video Task first and use only an intended disposable task. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
ID of the queued task to cancel or terminal task record to permanently delete.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_COUNT_MESSAGE_TOKENSCount input tokens for a BytePlus Ark Anthropic-compatible Messages request without generating a model response. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
BytePlus Ark model or inference endpoint ID whose tokenizer should be used.
array
Tool definitions whose framing should be counted.
number
Nucleus-sampling probability included in request framing.
string
System prompt to include in the token count.
array
required
Complete ordered conversation to include in the token count.
object
Request metadata included in request framing.
object
Extended-thinking configuration to include in the count.
integer
Maximum output tokens represented in request framing.
number
Sampling temperature from 0 through 2 included in request framing.
object
Tool-selection configuration to include in request framing.
string ("default" | "flex")
Online inference service tier included in request framing.
object
Output configuration included in request framing.
string
One stop sequence included in request framing.
array
System content blocks to include in the token count.
array
Up to four stop sequences included in request framing.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_CREATE_CHAT_COMPLETIONGenerate one synchronous chat completion from an activated BytePlus Ark model or inference endpoint in the connected Johor or Dublin region. Text, vision, video, audio, file, reasoning, and tool support varies by model. Streaming is intentionally unavailable because the action runner expects one JSON response.

Input Parameters

string
One model-supported stop sequence. Prefer stop_sequences when passing multiple values.
string
required
Activated model ID or inference endpoint ID in the connected region. Parameter and modality support varies by model.
array
Functions available for the model to call.
number
Nucleus sampling threshold from 0 through 1.
boolean
Whether to return output-token log probabilities.
array
required
Conversation history in chronological order.
object
Deep-reasoning configuration for supported models.
object
Token ID to bias mapping; values must be from -100 through 100.
integer
Maximum answer tokens; model limits vary.
number
Sampling temperature from 0 through 2.
string
Tool-selection policy: none, auto, or required. Use forced_function_name to require a specific function.
string ("fast" | "auto" | "default" | "flex")
Requested inference service tier.
integer
Most likely tokens to return per position (0-20).
array
One or more model-supported stop sequences.
object
Text, JSON object, or JSON Schema response configuration.
number
Presence penalty from -2 through 2.
string
Stable cache-routing identifier for related prompts.
string ("none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max")
Reasoning depth for supported models.
number
Frequency penalty from -2 through 2.
boolean
Whether the model may request multiple tools in one response.
string
Name of the specific available function the model must call.
integer
Maximum answer plus reasoning tokens; cannot be combined with max_tokens.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_CREATE_MESSAGEGenerate one complete, non-streaming message through BytePlus Ark’s official Anthropic-compatible Messages API. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
BytePlus Ark model or inference endpoint ID.
array
Anthropic-compatible tool definitions available to the model.
number
Nucleus-sampling probability from 0 through 1.
string
System prompt as text.
array
required
Complete ordered conversation history.
object
Request metadata accepted by the compatible Messages API.
object
Extended-thinking configuration supported by the model.
integer
Maximum output tokens to generate.
number
Sampling temperature from 0 through 2.
object
Anthropic-compatible tool-selection configuration.
string ("default" | "flex")
Online inference service tier.
object
Output configuration, including supported reasoning effort.
string
One text sequence that stops generation.
array
System prompt as supported content blocks; use system for plain text.
array
Up to four text sequences that stop generation.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_CREATE_MULTIMODAL_EMBEDDINGCreate one semantic embedding from mixed text, image, and supported video inputs using an activated skylark-embedding-vision model. Video input, sparse output, dimensions, and input combinations depend on the model version. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

array
required
Text, image, and model-supported video inputs embedded together.
string
required
Activated skylark-embedding-vision model ID or endpoint ID in the connected region. Video, dimensions, and mixed-input limits depend on model version.
integer ("1024" | "2048")
Dense vector dimension; supported by model versions 250615 and later.
string
Optional inference instruction for the embedding model.
string ("float" | "base64")
Return the dense vector as floats or Base64.
object
Sparse-vector output configuration; pure text input only.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_CREATE_VIDEO_TASKStart a potentially billable asynchronous video generation task and return its task ID. Model capabilities and optional parameter support vary. A callback URL, if supplied, may receive task-status POSTs. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

integer
Generation seed from -1 through 2,147,483,647; use -1 or omit it for provider-selected randomness. Supported by Seedance 1.5 pro and Seedance 1.0 models, not Seedance 2.5.
boolean
Whether to create a draft task; supported only by applicable models.
string
required
Video-generation model ID or configured inference endpoint ID available in the connected region. This free-form ID does not make this action validate model-specific parameter compatibility.
string ("16:9" | "4:3" | "1:1" | "3:4" | "9:16" | "21:9" | "adaptive")
Requested aspect ratio, such as 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, or adaptive; support depends on the model. Seedance 2.5 requires adaptive for editing, extension, and first-frame image modes.
integer
Requested frame count from 29 through 289 matching 25 + 4n; supported by Seedance 1.0 pro and pro fast, not Seedance 2.5. This value takes precedence over duration.
array
required
Ordered model inputs. Seedance 2.5 supports an optional text item plus up to 50 total references: at most 30 reference images, 10 reference videos, and 10 reference audio clips. Other models have lower limits and different supported combinations; this schema caps total items but does not infer compatibility from the free-form model ID.
integer
Requested video duration in seconds; allowed ranges are model-dependent. Seedance 2.5 supports 4-30 or -1, and editing requires -1. frames takes precedence when both are supplied.
integer
Queue priority from 0 through 9 for Seedance 2.5 and Seedance 2.0 models; larger values run first within the same endpoint. Not supported with flex.
boolean
Whether to add the provider watermark to the generated video.
string ("480p" | "720p" | "1080p")
Requested output resolution, commonly 480p, 720p, or 1080p; supported values and defaults depend on the model.
string
HTTPS endpoint to receive provider POST callbacks when task status changes. Omit it to avoid callback side effects.
boolean
Whether to request a fixed camera. Supported by Seedance 1.5 pro and Seedance 1.0 models, not Seedance 2.5, and unavailable for reference-image scenarios.
string ("default" | "flex")
Execution tier: default for online processing or flex for supported lower-cost offline processing. Seedance 2.5 and Seedance 2.0 models do not support flex.
string ("mp4" | "mov")
Seedance 2.5 output container: mp4 (default, broad compatibility) or mov (high color precision for professional workflows). Omit for other models.
boolean
Whether to generate audio with the video; omit this to use the selected model’s native default.
boolean
Whether the completed task should include a last-frame image URL.
string
Stable privacy-preserving end-user identifier of no more than 64 characters; BytePlus recommends English characters.
integer
Seconds the queued task may wait before expiring (3,600-259,200).
string ("reference" | "edit" | "extend" | "auto")
Seedance 2.5 omni-reference task hint. auto lets the model infer the task and can fail asynchronously for incompatible parameters. Explicit reference, edit, or extend enables earlier API constraint checks, but the task can still fail asynchronously if the prompt implies a different type. edit requires a 4-30 second reference video and uses ratio=adaptive and duration=-1 by default; extend requires a reference video and uses ratio=adaptive by default. Omit for other models.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_DELETE_FILEPermanently delete a BytePlus Ark file from storage. The file cannot be restored or used by later model inference requests. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
ID of the file to permanently delete.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_DELETE_RESPONSEPermanently delete one stored BytePlus Ark Responses API object by ID. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
Stored response ID to permanently delete.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_GENERATE_IMAGEGenerate or edit images synchronously with an activated Seedream model in the connected Johor or Dublin region. Image counts, reference limits, dimensions, prompt optimization, sequences, transparency, and layer decomposition are model-specific. Streaming is intentionally unavailable. This operation is billable.

Input Parameters

string
Resolution label or widthxheight; valid ranges depend on the Seedream model.
string
Accessible image URL or data URL for a single reference image.
string
required
Activated Seedream model ID or inference endpoint ID in the connected region. Size, editing, decomposition, and sequence support varies by model.
array
Model-supported content-generation tools such as web search.
array
Model-supported list of reference image URLs or data URLs.
string
Generation or editing prompt; optional only for automatic layer decomposition.
boolean
Whether to add the provider’s AI-generated watermark.
string ("transparent" | "opaque")
Background mode for supported single-image alpha-channel edits.
string ("png" | "jpeg")
Generated image file format for supported models.
boolean
Whether to apply provider prompt optimization.
string ("url" | "b64_json")
Return temporary URLs or Base64-encoded image data.
boolean
Decompose one image into a base image and layers; Seedream 5.0 pro/flash only.
object
Prompt optimization settings.
string ("auto" | "disabled")
Generate a related image sequence on Seedream 5.0 lite, 4.5, or 4.0.
object
Sequential generation limits.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_GET_FILERetrieve one BytePlus Ark file’s metadata, preprocessing status, expiration, and failure details by file ID. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
BytePlus Ark file ID returned by upload or list.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_GET_RESPONSERetrieve a stored BytePlus Ark Responses API object by ID, including its status, output, configuration, and token usage. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
Response ID returned by the Responses API.
boolean
Include source image URLs in returned message input-image items.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_GET_VIDEO_TASKRetrieve one video generation task’s status, failure details, and generated output URLs by task ID. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
required
Video generation task ID returned by create or list.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_LIST_FILESReturn one cursor-paginated page of BytePlus Ark files, optionally filtered by purpose or Managed Agents session scope. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
File ID cursor returned as next_cursor by a previous call. Omit it for the first page.
integer
Maximum files to return in this page (1-100).
string ("asc" | "desc")
Sort by creation time in ascending or descending order.
string ("user_data" | "agent")
Return only files with this purpose.
string
Managed Agents session ID whose mounted and generated files to return.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_LIST_RESPONSE_INPUT_ITEMSReturn one page of the input context associated with a BytePlus Ark response, with forward or backward ID pagination. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
Return items after this input item ID.
integer
Maximum items to return (1-100).
string ("asc" | "desc")
Sort order by item creation time.
string
Return items before this input item ID.
string
required
Response ID whose input context should be listed.
boolean
Include source image URLs in returned message input-image items.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_LIST_VIDEO_TASKSReturn one page of video generation tasks, optionally filtered by status, task IDs, model, or service tier. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
Return only tasks created with this exact model or inference endpoint ID.
string ("queued" | "running" | "cancelled" | "succeeded" | "failed")
Return only tasks with this status.
integer
Page number to return (1-500); increment it to request the next page.
array
Return only these task IDs. Each ID is sent as a separate repeated filter.task_ids query parameter.
integer
Maximum number of tasks to return in this page (1-500).
string ("default" | "flex")
Return only tasks using the default or flex service tier.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_TOKENIZE_TEXTTokenize text with an activated BytePlus Ark text model or inference endpoint This endpoint accepts text only; token vocabulary and model availability are model- and region-specific. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
A single text string to tokenize; use texts for multiple strings.
string
required
Activated text model ID or inference endpoint ID in the connected region. The tokenization endpoint currently supports text input only.
array
Multiple text strings to tokenize in one request; use text for one string.

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not
Slug: BYTE_PLUS_ARK_UPLOAD_FILEUpload a local file or import an HTTP(S) or BytePlus TOS file into BytePlus Ark for model or agent use. Files expire automatically; video preprocessing may continue after this request returns. Requires a Johor (ap-southeast-1) connection and is unavailable in Dublin (eu-west-1).

Input Parameters

string
Public HTTP(S) URL or BytePlus TOS URI. Use either url or file, not both.
object
A local file supplied for upload to BytePlus Ark.
string ("user_data" | "agent")
required
File purpose: user_data for general model input or agent for agent files.
integer
UTC Unix expiration timestamp. It must be 1-30 days from upload time; omit it for the provider’s 7-day default.
string
Existing BytePlus TOS bucket used to store the uploaded file.
string
Relative path prefix in the TOS bucket, such as arkfiles/.
number
Video frame sampling rate per second (0.2-5).
string
Model or inference endpoint ID used to choose video frame sampling defaults.
integer
Maximum video preprocessing token budget (10,240-204,800).
integer
Minimum sampled frame count for short videos (5-16).
integer
Maximum token budget per sampled frame (128-640).
integer
Minimum token budget per sampled frame (16-128).

Output

string
required
Data from the action execution
string
Error if any occurred during the execution of the action
boolean
required
Whether or not the action execution was successful or not