mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-18 04:44:37 +00:00
fix(volcengine): bill every Seedance video at create and keep its polls on the deployment that made it
Ark bills a task when it finishes while litellm bills it at create, so a request whose cost is not known then is rejected: Ark-picked length (duration -1), frames, a video input, and with it edit and extension. An omitted length is sent as 5s, since the 2.5 family otherwise lets Ark pick it and logged $0. Video ids now carry the requested and the served model group, read from litellm_metadata, so polls and downloads go back to the deployment that created the task. The routing and input_reference helpers move out of the DashScope config into a shared module both providers use. A download before the render finishes answers 400, unreadable or unmapped task bodies answer 502, malformed seconds, size, parameters and Ark fields raise unless drop_params is set, an unpriced resolution tier is refused, and the rendered frame table moves into the cost map. Co-Authored-By: AaronHowell <111272993+AaronHowell@users.noreply.github.com> Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Rick
parent
6d02f5862f
commit
eb3cb3b192
9 changed files with 1256 additions and 1636 deletions
106
litellm/llms/base_llm/videos/async_task_utils.py
Normal file
106
litellm/llms/base_llm/videos/async_task_utils.py
Normal file
|
|
@ -0,0 +1,106 @@
|
|||
import base64
|
||||
from collections.abc import Mapping
|
||||
from io import BufferedReader, BytesIO
|
||||
from pathlib import Path, PurePath
|
||||
from typing import Final
|
||||
|
||||
from pydantic import TypeAdapter, ValidationError
|
||||
|
||||
from litellm.images.utils import ImageEditRequestUtils
|
||||
from litellm.types.videos.utils import decode_video_id_with_provider
|
||||
|
||||
_FIELDS_ADAPTER: Final = TypeAdapter(dict[str, object])
|
||||
_FILE_TUPLE_ADAPTER: Final = TypeAdapter(tuple[object, ...])
|
||||
|
||||
|
||||
def _fields(value: object) -> Mapping[str, object] | None:
|
||||
try:
|
||||
return _FIELDS_ADAPTER.validate_python(value)
|
||||
except ValidationError:
|
||||
return None
|
||||
|
||||
|
||||
def _logged_litellm_params(logging_obj: object) -> Mapping[str, object]:
|
||||
return _fields(getattr(logging_obj, "litellm_params", None)) or {}
|
||||
|
||||
|
||||
def _id_safe_name(name: object) -> str | None:
|
||||
return name if isinstance(name, str) and name and ";" not in name else None
|
||||
|
||||
|
||||
def _stamped_models(metadata: object) -> tuple[str, str | None] | None:
|
||||
fields: Final = _fields(metadata)
|
||||
if fields is None:
|
||||
return None
|
||||
requested: Final = _id_safe_name(fields.get("model_group"))
|
||||
served: Final = _id_safe_name(fields.get("deployment_model_name"))
|
||||
if requested is None:
|
||||
return None
|
||||
return requested, None if served is None or served == requested or "*" in served else served
|
||||
|
||||
|
||||
def routed_video_models(logging_obj: object) -> tuple[str, str | None] | None:
|
||||
"""
|
||||
A later status or content request checks the key's model access against the model encoded in the id, and
|
||||
routes by it. One upstream model name can be served by several deployments (regions, accounts), so only the
|
||||
group that served the task can send its polls back to the deployment that created it, while a key may be
|
||||
scoped to the alias, routing group or wildcard name it asked for instead. The id carries both names: the one
|
||||
the router was asked for on this hop, and the group that served it when that differs and is not a wildcard.
|
||||
The router stamps both into litellm_metadata, and metadata carries a copy only when the caller sent none.
|
||||
"""
|
||||
params: Final = _logged_litellm_params(logging_obj)
|
||||
stamped: Final = (_stamped_models(params.get(key)) for key in ("litellm_metadata", "metadata"))
|
||||
return next((models for models in stamped if models), None)
|
||||
|
||||
|
||||
def polled_video_models(logging_obj: object) -> tuple[str | None, str | None]:
|
||||
"""
|
||||
The served group survives into the returned id only when the router sent this poll to it, so an edited id
|
||||
cannot write any name it likes back.
|
||||
"""
|
||||
video_id: Final = _logged_litellm_params(logging_obj).get("video_id")
|
||||
if not isinstance(video_id, str):
|
||||
return None, None
|
||||
decoded: Final = decode_video_id_with_provider(video_id)
|
||||
served: Final = decoded.get("served_model")
|
||||
routed: Final = routed_video_models(logging_obj)
|
||||
return decoded.get("model_id") or None, served if routed is None or served in routed else None
|
||||
|
||||
|
||||
def _file_content(file_obj: object) -> object:
|
||||
if not isinstance(file_obj, tuple):
|
||||
return file_obj
|
||||
file_tuple: Final = _FILE_TUPLE_ADAPTER.validate_python(file_obj)
|
||||
return file_tuple[1] if len(file_tuple) >= 2 else None
|
||||
|
||||
|
||||
def input_reference_bytes(file_obj: object) -> bytes:
|
||||
content: Final = _file_content(file_obj)
|
||||
if isinstance(content, PurePath):
|
||||
return Path(content).read_bytes()
|
||||
if isinstance(content, (BytesIO, BufferedReader)):
|
||||
current_position: Final = content.tell()
|
||||
_ = content.seek(0)
|
||||
data: Final = content.read()
|
||||
_ = content.seek(current_position)
|
||||
return data
|
||||
if isinstance(content, bytes):
|
||||
return content
|
||||
if isinstance(content, bytearray):
|
||||
return bytes(content)
|
||||
read: Final = getattr(content, "read", None)
|
||||
if callable(read):
|
||||
read_data: Final[object] = read()
|
||||
if isinstance(read_data, bytes):
|
||||
return read_data
|
||||
raise ValueError("input_reference must be a URL string, bytes, a path, a file object, or a file tuple")
|
||||
|
||||
|
||||
def input_reference_url(image: object) -> str:
|
||||
"""A string is a URL the provider fetches itself and is never opened as a local path."""
|
||||
if isinstance(image, str):
|
||||
return image
|
||||
content: Final = input_reference_bytes(image)
|
||||
content_type: Final = ImageEditRequestUtils.get_image_content_type(content)
|
||||
encoded: Final = base64.b64encode(content).decode("utf-8")
|
||||
return f"data:{content_type};base64,{encoded}"
|
||||
|
|
@ -3,13 +3,10 @@ DashScope (Alibaba Model Studio) async video task API: create, poll and download
|
|||
delete or remix endpoint, so those raise.
|
||||
"""
|
||||
|
||||
import base64
|
||||
import json
|
||||
from collections.abc import Mapping
|
||||
from datetime import datetime
|
||||
from io import BufferedReader, BytesIO
|
||||
from math import log
|
||||
from pathlib import Path, PurePath
|
||||
from types import MappingProxyType
|
||||
from typing import TYPE_CHECKING, Annotated, Final
|
||||
|
||||
|
|
@ -19,9 +16,13 @@ from pydantic import BaseModel, Field, StrictBool, StrictInt, StrictStr, TypeAda
|
|||
|
||||
import litellm
|
||||
from litellm.exceptions import UnsupportedParamsError
|
||||
from litellm.images.utils import ImageEditRequestUtils
|
||||
from litellm.litellm_core_utils.url_utils import encode_url_path_segment
|
||||
from litellm.llms.base_llm.chat.transformation import BaseLLMException
|
||||
from litellm.llms.base_llm.videos.async_task_utils import (
|
||||
input_reference_url,
|
||||
polled_video_models,
|
||||
routed_video_models,
|
||||
)
|
||||
from litellm.llms.base_llm.videos.transformation import BaseVideoConfig
|
||||
from litellm.llms.custom_httpx.http_handler import (
|
||||
_get_httpx_client, # pyright: ignore[reportPrivateUsage, reportUnknownVariableType] # house cached-client factory has no public alias and its stub leaves params untyped
|
||||
|
|
@ -32,7 +33,6 @@ from litellm.types.router import GenericLiteLLMParams
|
|||
from litellm.types.utils import ModelInfo
|
||||
from litellm.types.videos.main import VideoCreateOptionalRequestParams, VideoObject
|
||||
from litellm.types.videos.utils import (
|
||||
decode_video_id_with_provider,
|
||||
encode_video_id_with_provider,
|
||||
extract_original_video_id,
|
||||
)
|
||||
|
|
@ -86,7 +86,6 @@ _EMPTY_PARAMS: Final[dict[str, object]] = {} # mutable-ok: BaseVideoConfig cont
|
|||
|
||||
_TASK_RESPONSE_ADAPTER: Final = TypeAdapter(_TaskResponse)
|
||||
_PARAMETERS_ADAPTER: Final = TypeAdapter(dict[str, object])
|
||||
_FILE_TUPLE_ADAPTER: Final = TypeAdapter(tuple[object, ...])
|
||||
|
||||
_STATUS_MAP: Final = MappingProxyType(
|
||||
{
|
||||
|
|
@ -383,39 +382,6 @@ def _is_priced_tier(resolution: object, priced_tiers: tuple[str, ...]) -> bool:
|
|||
return isinstance(resolution, str) and resolution.strip().upper() in priced_tiers
|
||||
|
||||
|
||||
def _read_all_bytes(file_obj: object) -> bytes:
|
||||
if isinstance(file_obj, tuple):
|
||||
file_tuple: Final = _FILE_TUPLE_ADAPTER.validate_python(file_obj)
|
||||
return _read_all_bytes(file_tuple[1] if len(file_tuple) >= 2 else None)
|
||||
if isinstance(file_obj, PurePath):
|
||||
return Path(file_obj).read_bytes()
|
||||
if isinstance(file_obj, (BytesIO, BufferedReader)):
|
||||
current_position: Final = file_obj.tell()
|
||||
file_obj.seek(0)
|
||||
content: Final = file_obj.read()
|
||||
file_obj.seek(current_position)
|
||||
return content
|
||||
if isinstance(file_obj, bytes):
|
||||
return file_obj
|
||||
if isinstance(file_obj, bytearray):
|
||||
return bytes(file_obj)
|
||||
read: Final = getattr(file_obj, "read", None)
|
||||
if callable(read):
|
||||
data: Final = read()
|
||||
if isinstance(data, bytes):
|
||||
return data
|
||||
raise ValueError("input_reference must be a URL string, bytes, a path, a file object, or a file tuple")
|
||||
|
||||
|
||||
def _image_url(image: object) -> str:
|
||||
if isinstance(image, str):
|
||||
return image
|
||||
content: Final = _read_all_bytes(image)
|
||||
content_type: Final = ImageEditRequestUtils.get_image_content_type(content)
|
||||
encoded: Final = base64.b64encode(content).decode("utf-8")
|
||||
return f"data:{content_type};base64,{encoded}"
|
||||
|
||||
|
||||
def _media_reference_type(model: str) -> str | None:
|
||||
"""Models without the field predate the media array and take the image on the flat ``img_url`` field."""
|
||||
info: Final = _dashscope_model_info(model)
|
||||
|
|
@ -558,58 +524,6 @@ def _video_url_from_task(task: _TaskResponse, raw_response: httpx.Response) -> s
|
|||
raise _video_error(raw_response, 502, "Video URL not found in task response. The task may not have succeeded yet.")
|
||||
|
||||
|
||||
def _logged_litellm_params(logging_obj: object) -> Mapping[str, object]:
|
||||
litellm_params: Final[object] = getattr(logging_obj, "litellm_params", None)
|
||||
try:
|
||||
return _PARAMETERS_ADAPTER.validate_python(litellm_params)
|
||||
except ValidationError:
|
||||
return _EMPTY_PARAMS
|
||||
|
||||
|
||||
def _polled_models(logging_obj: object) -> tuple[str | None, str | None]:
|
||||
"""
|
||||
The proxy routes by the models encoded in the polled id, so they must survive into the returned id. The served
|
||||
group survives only when the router sent this poll to it, so an edited id cannot write any name it likes back.
|
||||
"""
|
||||
video_id: Final = _logged_litellm_params(logging_obj).get("video_id")
|
||||
if not isinstance(video_id, str):
|
||||
return None, None
|
||||
decoded: Final = decode_video_id_with_provider(video_id)
|
||||
served: Final = decoded.get("served_model")
|
||||
routed: Final = _poll_models(logging_obj)
|
||||
return decoded.get("model_id") or None, served if routed is None or served in routed else None
|
||||
|
||||
|
||||
def _poll_models(logging_obj: object) -> tuple[str, str | None] | None:
|
||||
"""
|
||||
A later status or content request checks the key's model access against the model encoded in the id, and
|
||||
routes by it. DashScope serves the same model name from several regional hosts, so only the group that
|
||||
served the task can send its polls back to the region that created it, while a key may be scoped to the
|
||||
alias, routing group or wildcard name it asked for instead. The id carries both names: the one the router
|
||||
was asked for on this hop, and the group that served it when that differs and is not a wildcard. The router
|
||||
stamps both into litellm_metadata, and metadata carries a copy only when the caller sent none.
|
||||
"""
|
||||
params: Final = _logged_litellm_params(logging_obj)
|
||||
stamped: Final = (_stamped_poll_models(params.get(key)) for key in ("litellm_metadata", "metadata"))
|
||||
return next((models for models in stamped if models), None)
|
||||
|
||||
|
||||
def _stamped_poll_models(metadata: object) -> tuple[str, str | None] | None:
|
||||
try:
|
||||
fields: Final = _PARAMETERS_ADAPTER.validate_python(metadata)
|
||||
except ValidationError:
|
||||
return None
|
||||
requested: Final = _id_safe_name(fields.get("model_group"))
|
||||
served: Final = _id_safe_name(fields.get("deployment_model_name"))
|
||||
if requested is None:
|
||||
return None
|
||||
return requested, None if served is None or served == requested or "*" in served else served
|
||||
|
||||
|
||||
def _id_safe_name(name: object) -> str | None:
|
||||
return name if isinstance(name, str) and name and ";" not in name else None
|
||||
|
||||
|
||||
class DashScopeVideoConfig(BaseVideoConfig):
|
||||
def get_supported_openai_params(self, model: str) -> list[str]: # mutable-ok: BaseVideoConfig contract returns list
|
||||
return [
|
||||
|
|
@ -804,7 +718,7 @@ class DashScopeVideoConfig(BaseVideoConfig):
|
|||
input_reference: Final = video_create_optional_request_params.get("input_reference")
|
||||
if input_reference is None:
|
||||
return None, None
|
||||
image_url: Final = _image_url(input_reference)
|
||||
image_url: Final = input_reference_url(input_reference)
|
||||
media_type: Final = _media_reference_type(model)
|
||||
if media_type is None:
|
||||
return "img_url", image_url
|
||||
|
|
@ -823,7 +737,7 @@ class DashScopeVideoConfig(BaseVideoConfig):
|
|||
requested: Final = self._requested_parameters(request_data)
|
||||
video_obj: Final = _video_object_from_task(task, raw_response, model=model, requested=requested)
|
||||
if custom_llm_provider and video_obj.id:
|
||||
requested_model, served_model = _poll_models(logging_obj) or (model, None)
|
||||
requested_model, served_model = routed_video_models(logging_obj) or (model, None)
|
||||
video_obj.id = encode_video_id_with_provider(
|
||||
video_obj.id, custom_llm_provider, requested_model, served_model
|
||||
)
|
||||
|
|
@ -869,7 +783,7 @@ class DashScopeVideoConfig(BaseVideoConfig):
|
|||
client: "HTTPHandler | None" = None,
|
||||
) -> VideoObject:
|
||||
task: Final = _parse_task_response(raw_response)
|
||||
model, served_model = _polled_models(logging_obj)
|
||||
model, served_model = polled_video_models(logging_obj)
|
||||
video_obj: Final = _video_object_from_task(task, raw_response, model=model)
|
||||
if custom_llm_provider and video_obj.id:
|
||||
video_obj.id = encode_video_id_with_provider(video_obj.id, custom_llm_provider, model, served_model)
|
||||
|
|
|
|||
|
|
@ -13,7 +13,6 @@ from .common_utils import (
|
|||
)
|
||||
from .embedding import VolcEngineEmbeddingConfig
|
||||
from .responses.transformation import VolcEngineResponsesAPIConfig
|
||||
from .videos.transformation import VolcEngineVideoConfig
|
||||
|
||||
# For backward compatibility, keep the old class name
|
||||
VolcEngineConfig: Final = VolcEngineChatConfig
|
||||
|
|
@ -24,7 +23,6 @@ __all__ = [
|
|||
"VolcEngineEmbeddingConfig",
|
||||
"VolcEngineError",
|
||||
"VolcEngineResponsesAPIConfig",
|
||||
"VolcEngineVideoConfig",
|
||||
"get_volcengine_base_url",
|
||||
"get_volcengine_headers",
|
||||
]
|
||||
|
|
|
|||
|
|
@ -1 +0,0 @@
|
|||
# Volcengine Ark video generation (Doubao-Seedance family, async task API)
|
||||
File diff suppressed because it is too large
Load diff
|
|
@ -59477,7 +59477,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s). List 70 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 42 CNY per million tokens with a video input. Accepts up to 50 reference assets (30 images, 10 videos, 10 audios) and clips of 4-30s; per-asset billing beyond the free allowance is not encoded. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "854x480", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "480x854", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s). List 70 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Accepts up to 50 reference assets (30 images, 10 videos, 10 audios) and clips of 4-30s; per-asset billing beyond the free allowance is not encoded. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-260128": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59491,7 +59492,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 46 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 28 CNY per million tokens with a video input. Clips of 4-15s; 4k is exclusive to this tier. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 46 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s; 4k is exclusive to this tier. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-fast-260128": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59503,7 +59505,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 37 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 22 CNY per million tokens with a video input. Clips of 4-15s at 480p or 720p. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 37 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s at 480p or 720p. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-mini-260615": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59515,7 +59518,8 @@
|
|||
"supported_modalities": ["text", "image"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 23 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 14 CNY per million tokens with a video input. Clips of 4-15s at 480p or 720p. Only text and image inputs are declared; the mini tier's reference video and audio caps were not verified. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 23 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s at 480p or 720p. Only text and image inputs are declared; the mini tier's reference video and audio caps were not verified. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"bedrock/us-east-1/zai.glm-5": {
|
||||
"input_cost_per_token": 1e-06,
|
||||
|
|
|
|||
|
|
@ -59477,7 +59477,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s). List 70 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 42 CNY per million tokens with a video input. Accepts up to 50 reference assets (30 images, 10 videos, 10 audios) and clips of 4-30s; per-asset billing beyond the free allowance is not encoded. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "854x480", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "480x854", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s). List 70 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Accepts up to 50 reference assets (30 images, 10 videos, 10 audios) and clips of 4-30s; per-asset billing beyond the free allowance is not encoded. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-260128": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59491,7 +59492,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 46 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 28 CNY per million tokens with a video input. Clips of 4-15s; 4k is exclusive to this tier. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 46 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s; 4k is exclusive to this tier. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-fast-260128": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59503,7 +59505,8 @@
|
|||
"supported_modalities": ["text", "image", "video", "audio"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 37 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 22 CNY per million tokens with a video input. Clips of 4-15s at 480p or 720p. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 37 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s at 480p or 720p. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"volcengine/doubao-seedance-2-0-mini-260615": {
|
||||
"litellm_provider": "volcengine",
|
||||
|
|
@ -59515,7 +59518,8 @@
|
|||
"supported_modalities": ["text", "image"],
|
||||
"supported_output_modalities": ["video"],
|
||||
"supported_endpoints": ["/v1/videos"],
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 23 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. 14 CNY per million tokens with a video input. Clips of 4-15s at 480p or 720p. Only text and image inputs are declared; the mini tier's reference video and audio caps were not verified. Requests that do carry a video input bill at the lower CNY rate instead, which litellm's per-second tiers cannot express, so edit, extension and reference-video jobs are overestimated."}
|
||||
"provider_specific_entry": {"volcengine_video_frames": {"480p": {"16:9": "864x496", "4:3": "752x560", "1:1": "640x640", "3:4": "560x752", "9:16": "496x864", "21:9": "992x432"}, "720p": {"16:9": "1280x720", "4:3": "1112x834", "1:1": "960x960", "3:4": "834x1112", "9:16": "720x1280", "21:9": "1470x630"}}},
|
||||
"metadata": {"comment": "Ark bills per output token, not per second: tokens = width * height * 24 * seconds / 1024, so a per-second rate per resolution tier is exact for the 16:9 geometry Ark publishes (1280x720 is 21600 tokens/s; the 480p frame is 864x496, 10044 tokens/s, not the tier's nominal 854x480). List 23 CNY per million tokens without a video input, converted at 7.0 CNY/USD like the other volcengine entries. Clips of 4-15s at 480p or 720p. Only text and image inputs are declared; the mini tier's reference video and audio caps were not verified. Requests with a video input bill at a lower rate per token and are rejected, since their cost is not known at create time. provider_specific_entry.volcengine_video_frames is the frame Ark renders per resolution tier and ratio, verified with ffprobe at 480p for every ratio and at 720p for 16:9 and 4:3; the other 720p ratios are from the Seedance 2.0 model card."}
|
||||
},
|
||||
"bedrock/us-east-1/zai.glm-5": {
|
||||
"input_cost_per_token": 1e-06,
|
||||
|
|
|
|||
|
|
@ -2748,7 +2748,8 @@
|
|||
"batches": false,
|
||||
"rerank": false,
|
||||
"a2a": true,
|
||||
"interactions": true
|
||||
"interactions": true,
|
||||
"video_generations": true
|
||||
}
|
||||
},
|
||||
"voyage": {
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load diff
Loading…
Add table
Reference in a new issue