docling 2.40.0__tar.gz → 2.42.0__tar.gz

{docling-2.40.0 → docling-2.42.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docling
-Version: 2.40.0
+Version: 2.42.0
 Summary: SDK and CLI for parsing PDF, DOCX, HTML, and more, to a unified document representation for powering downstream workflows such as gen AI applications.
 Author-email: Christoph Auer <cau@zurich.ibm.com>, Michele Dolfi <dol@zurich.ibm.com>, Maxim Lysak <mly@zurich.ibm.com>, Nikos Livathinos <nli@zurich.ibm.com>, Ahmed Nassar <ahn@zurich.ibm.com>, Panos Vagenas <pva@zurich.ibm.com>, Peter Staar <taa@zurich.ibm.com>
 License-Expression: MIT
@@ -26,7 +26,7 @@ Requires-Python: <4.0,>=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: pydantic<3.0.0,>=2.0.0
-Requires-Dist: docling-core[chunking]<3.0.0,>=2.39.0
+Requires-Dist: docling-core[chunking]<3.0.0,>=2.42.0
 Requires-Dist: docling-parse<5.0.0,>=4.0.0
 Requires-Dist: docling-ibm-models<4,>=3.6.0
 Requires-Dist: filetype<2.0.0,>=1.2.0
@@ -50,6 +50,7 @@ Requires-Dist: tqdm<5.0.0,>=4.65.0
 Requires-Dist: pluggy<2.0.0,>=1.0.0
 Requires-Dist: pylatexenc<3.0,>=2.10
 Requires-Dist: scipy<2.0.0,>=1.6.0
+Requires-Dist: accelerate<2,>=1.0.0
 Provides-Extra: tesserocr
 Requires-Dist: tesserocr<3.0.0,>=2.7.1; extra == "tesserocr"
 Provides-Extra: ocrmac
@@ -62,7 +63,7 @@ Provides-Extra: rapidocr
 Requires-Dist: rapidocr-onnxruntime<2.0.0,>=1.4.0; python_version < "3.13" and extra == "rapidocr"
 Requires-Dist: onnxruntime<2.0.0,>=1.7.0; extra == "rapidocr"
 Provides-Extra: asr
-Requires-Dist: openai-whisper>=20240930; extra == "asr"
+Requires-Dist: openai-whisper>=20250625; extra == "asr"
 Dynamic: license-file
 
 <p align="center">

{docling-2.40.0 → docling-2.42.0}/docling/backend/docx/latex/omml.py

@@ -260,7 +260,15 @@ class oMath2Latex(Tag2Method):
         the fraction object
         """
         c_dict = self.process_children_dict(elm)
-        pr = c_dict["fPr"]
+        pr = c_dict.get("fPr")
+        if pr is None:
+            # Handle missing fPr element gracefully
+            _log.debug("Missing fPr element in fraction, using default formatting")
+            latex_s = F_DEFAULT
+            return latex_s.format(
+                num=c_dict.get("num"),
+                den=c_dict.get("den"),
+            )
         latex_s = get_val(pr.type, default=F_DEFAULT, store=F)
         return pr.text + latex_s.format(num=c_dict.get("num"), den=c_dict.get("den"))
 

{docling-2.40.0 → docling-2.42.0}/docling/backend/html_backend.py

@@ -379,6 +379,25 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
         else:
             _log.debug(f"list-item has no text: {element}")
 
+    @staticmethod
+    def _get_cell_spans(cell: Tag) -> tuple[int, int]:
+        """Extract colspan and rowspan values from a table cell tag.
+
+        This function retrieves the 'colspan' and 'rowspan' attributes from a given
+        table cell tag.
+        If the attribute does not exist or it is not numeric, it defaults to 1.
+        """
+        raw_spans: tuple[str, str] = (
+            str(cell.get("colspan", "1")),
+            str(cell.get("rowspan", "1")),
+        )
+        int_spans: tuple[int, int] = (
+            int(raw_spans[0]) if raw_spans[0].isnumeric() else 1,
+            int(raw_spans[1]) if raw_spans[0].isnumeric() else 1,
+        )
+
+        return int_spans
+
     @staticmethod
     def parse_table_data(element: Tag) -> Optional[TableData]:  # noqa: C901
         nested_tables = element.find("table")
@@ -398,10 +417,9 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
                 if not isinstance(row, Tag):
                     continue
                 cell_tag = cast(Tag, cell)
-                val = cell_tag.get("colspan", "1")
-                colspan = int(val) if (isinstance(val, str) and val.isnumeric()) else 1
-                col_count += colspan
-                if cell_tag.name == "td" or cell_tag.get("rowspan") is None:
+                col_span, row_span = HTMLDocumentBackend._get_cell_spans(cell_tag)
+                col_count += col_span
+                if cell_tag.name == "td" or row_span == 1:
                     is_row_header = False
             num_cols = max(num_cols, col_count)
             if not is_row_header:
@@ -428,10 +446,11 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
             row_header = True
             for html_cell in cells:
                 if isinstance(html_cell, Tag):
+                    _, row_span = HTMLDocumentBackend._get_cell_spans(html_cell)
                     if html_cell.name == "td":
                         col_header = False
                         row_header = False
-                    elif html_cell.get("rowspan") is None:
+                    elif row_span == 1:
                         row_header = False
             if not row_header:
                 row_idx += 1
@@ -456,18 +475,7 @@ class HTMLDocumentBackend(DeclarativeDocumentBackend):
                 text = html_cell.text
 
                 # label = html_cell.name
-                col_val = html_cell.get("colspan", "1")
-                col_span = (
-                    int(col_val)
-                    if isinstance(col_val, str) and col_val.isnumeric()
-                    else 1
-                )
-                row_val = html_cell.get("rowspan", "1")
-                row_span = (
-                    int(row_val)
-                    if isinstance(row_val, str) and row_val.isnumeric()
-                    else 1
-                )
+                col_span, row_span = HTMLDocumentBackend._get_cell_spans(html_cell)
                 if row_header:
                     row_span -= 1
                 while (

{docling-2.40.0 → docling-2.42.0}/docling/backend/xml/jats_backend.py

@@ -93,8 +93,8 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
 
         # Initialize the root of the document hierarchy
         self.root: Optional[NodeItem] = None
-
-        self.valid = False
+        self.hlevel: int = 0
+        self.valid: bool = False
         try:
             if isinstance(self.path_or_stream, BytesIO):
                 self.path_or_stream.seek(0)
@@ -147,6 +147,7 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 binary_hash=self.document_hash,
             )
             doc = DoclingDocument(name=self.file.stem or "file", origin=origin)
+            self.hlevel = 0
 
             # Get metadata XML components
             xml_components: XMLComponents = self._parse_metadata()
@@ -304,7 +305,9 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
             title: str = abstract["label"] or DEFAULT_HEADER_ABSTRACT
             if not text:
                 continue
-            parent = doc.add_heading(parent=self.root, text=title)
+            parent = doc.add_heading(
+                parent=self.root, text=title, level=self.hlevel + 1
+            )
             doc.add_text(
                 parent=parent,
                 text=text,
@@ -637,7 +640,10 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 elif child.tag == "ack":
                     text = DEFAULT_HEADER_ACKNOWLEDGMENTS
                 if text:
-                    new_parent = doc.add_heading(text=text, parent=parent)
+                    self.hlevel += 1
+                    new_parent = doc.add_heading(
+                        text=text, parent=parent, level=self.hlevel
+                    )
             elif child.tag == "list":
                 new_parent = doc.add_group(
                     label=GroupLabel.LIST, name="list", parent=parent
@@ -694,6 +700,8 @@ class JatsDocumentBackend(DeclarativeDocumentBackend):
                 new_text = self._walk_linear(doc, new_parent, child)
                 if not (node.getparent().tag == "p" and node.tag in flush_tags):
                     node_text += new_text
+                if child.tag in ("sec", "ack") and text:
+                    self.hlevel -= 1
 
             # pick up the tail text
             node_text += child.tail.replace("\n", " ") if child.tail else ""

{docling-2.40.0 → docling-2.42.0}/docling/datamodel/asr_model_specs.py

@@ -22,7 +22,7 @@ WHISPER_TINY = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )
@@ -33,7 +33,7 @@ WHISPER_SMALL = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )
@@ -44,7 +44,7 @@ WHISPER_MEDIUM = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )
@@ -55,7 +55,7 @@ WHISPER_BASE = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )
@@ -66,7 +66,7 @@ WHISPER_LARGE = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )
@@ -77,7 +77,7 @@ WHISPER_TURBO = InlineAsrNativeWhisperOptions(
     verbose=True,
     timestamps=True,
     word_timestamps=True,
-    temperatue=0.0,
+    temperature=0.0,
     max_new_tokens=256,
     max_time_chunk=30.0,
 )

{docling-2.40.0 → docling-2.42.0}/docling/datamodel/base_models.py

@@ -12,6 +12,7 @@ from docling_core.types.doc import (
     Size,
     TableCell,
 )
+from docling_core.types.doc.base import PydanticSerCtxKey, round_pydantic_float
 from docling_core.types.doc.page import SegmentedPdfPage, TextCell
 from docling_core.types.io import (
     DocumentStream,
@@ -19,7 +20,14 @@ from docling_core.types.io import (
 
 # DO NOT REMOVE; explicitly exposed from this location
 from PIL.Image import Image
-from pydantic import BaseModel, ConfigDict, Field, computed_field
+from pydantic import (
+    BaseModel,
+    ConfigDict,
+    Field,
+    FieldSerializationInfo,
+    computed_field,
+    field_serializer,
+)
 
 if TYPE_CHECKING:
     from docling.backend.pdf_backend import PdfPageBackend
@@ -142,6 +150,10 @@ class Cluster(BaseModel):
     cells: List[TextCell] = []
     children: List["Cluster"] = []  # Add child cluster support
 
+    @field_serializer("confidence")
+    def _serialize(self, value: float, info: FieldSerializationInfo) -> float:
+        return round_pydantic_float(value, info.context, PydanticSerCtxKey.CONFID_PREC)
+
 
 class BasePageElement(BaseModel):
     label: DocItemLabel
@@ -194,6 +206,16 @@ class FigureElement(BasePageElement):
     predicted_class: Optional[str] = None
     confidence: Optional[float] = None
 
+    @field_serializer("confidence")
+    def _serialize(
+        self, value: Optional[float], info: FieldSerializationInfo
+    ) -> Optional[float]:
+        return (
+            round_pydantic_float(value, info.context, PydanticSerCtxKey.CONFID_PREC)
+            if value is not None
+            else None
+        )
+
 
 class FigureClassificationPrediction(BaseModel):
     figure_count: int = 0

docling-2.42.0/docling/datamodel/layout_model_specs.py

@@ -0,0 +1,90 @@
+import logging
+from enum import Enum
+from pathlib import Path
+from typing import Optional
+
+from pydantic import BaseModel
+
+from docling.datamodel.accelerator_options import AcceleratorDevice
+
+_log = logging.getLogger(__name__)
+
+
+class LayoutModelConfig(BaseModel):
+    name: str
+    repo_id: str
+    revision: str
+    model_path: str
+    supported_devices: list[AcceleratorDevice] = [
+        AcceleratorDevice.CPU,
+        AcceleratorDevice.CUDA,
+        AcceleratorDevice.MPS,
+    ]
+
+    @property
+    def model_repo_folder(self) -> str:
+        return self.repo_id.replace("/", "--")
+
+
+# HuggingFace Layout Models
+
+# Default Docling Layout Model
+DOCLING_LAYOUT_V2 = LayoutModelConfig(
+    name="docling_layout_v2",
+    repo_id="ds4sd/docling-layout-old",
+    revision="main",
+    model_path="",
+)
+
+DOCLING_LAYOUT_HERON = LayoutModelConfig(
+    name="docling_layout_heron",
+    repo_id="ds4sd/docling-layout-heron",
+    revision="main",
+    model_path="",
+)
+
+DOCLING_LAYOUT_HERON_101 = LayoutModelConfig(
+    name="docling_layout_heron_101",
+    repo_id="ds4sd/docling-layout-heron-101",
+    revision="main",
+    model_path="",
+)
+
+DOCLING_LAYOUT_EGRET_MEDIUM = LayoutModelConfig(
+    name="docling_layout_egret_medium",
+    repo_id="ds4sd/docling-layout-egret-medium",
+    revision="main",
+    model_path="",
+)
+
+DOCLING_LAYOUT_EGRET_LARGE = LayoutModelConfig(
+    name="docling_layout_egret_large",
+    repo_id="ds4sd/docling-layout-egret-large",
+    revision="main",
+    model_path="",
+)
+
+DOCLING_LAYOUT_EGRET_XLARGE = LayoutModelConfig(
+    name="docling_layout_egret_xlarge",
+    repo_id="ds4sd/docling-layout-egret-xlarge",
+    revision="main",
+    model_path="",
+)
+
+# Example for a hypothetical alternative model
+# ALTERNATIVE_LAYOUT = LayoutModelConfig(
+#     name="alternative_layout",
+#     repo_id="someorg/alternative-layout",
+#     revision="main",
+#     model_path="model_artifacts/layout_alt",
+# )
+
+
+class LayoutModelType(str, Enum):
+    DOCLING_LAYOUT_V2 = "docling_layout_v2"
+    DOCLING_LAYOUT_HERON = "docling_layout_heron"
+    DOCLING_LAYOUT_HERON_101 = "docling_layout_heron_101"
+    DOCLING_LAYOUT_EGRET_MEDIUM = "docling_layout_egret_medium"
+    DOCLING_LAYOUT_EGRET_LARGE = "docling_layout_egret_large"
+    DOCLING_LAYOUT_EGRET_XLARGE = "docling_layout_egret_xlarge"
+    # ALTERNATIVE_LAYOUT = "alternative_layout"

{docling-2.40.0 → docling-2.42.0}/docling/datamodel/pipeline_options.py

@@ -16,6 +16,15 @@ from docling.datamodel import asr_model_specs
 
 # Import the following for backwards compatibility
 from docling.datamodel.accelerator_options import AcceleratorDevice, AcceleratorOptions
+from docling.datamodel.layout_model_specs import (
+    DOCLING_LAYOUT_EGRET_LARGE,
+    DOCLING_LAYOUT_EGRET_MEDIUM,
+    DOCLING_LAYOUT_EGRET_XLARGE,
+    DOCLING_LAYOUT_HERON,
+    DOCLING_LAYOUT_HERON_101,
+    DOCLING_LAYOUT_V2,
+    LayoutModelConfig,
+)
 from docling.datamodel.pipeline_options_asr_model import (
     InlineAsrOptions,
 )
@@ -208,7 +217,7 @@ smolvlm_picture_description = PictureDescriptionVlmOptions(
 
 # GraniteVision
 granite_picture_description = PictureDescriptionVlmOptions(
-    repo_id="ibm-granite/granite-vision-3.2-2b-preview",
+    repo_id="ibm-granite/granite-vision-3.3-2b",
     prompt="What is shown in this image?",
 )
 
@@ -270,6 +279,10 @@ class LayoutOptions(BaseModel):
     """Options for layout processing."""
 
     create_orphan_clusters: bool = True  # Whether to create clusters for orphaned cells
+    keep_empty_clusters: bool = (
+        False  # Whether to keep clusters that contain no text cells
+    )
+    model_spec: LayoutModelConfig = DOCLING_LAYOUT_V2
 
 
 class AsrPipelineOptions(PipelineOptions):

{docling-2.40.0 → docling-2.42.0}/docling/datamodel/pipeline_options_vlm_model.py

@@ -1,6 +1,7 @@
 from enum import Enum
-from typing import Any, Dict, List, Literal, Optional, Union
+from typing import Any, Callable, Dict, List, Literal, Optional, Union
 
+from docling_core.types.doc.page import SegmentedPage
 from pydantic import AnyUrl, BaseModel
 from typing_extensions import deprecated
 
@@ -9,9 +10,10 @@ from docling.datamodel.accelerator_options import AcceleratorDevice
 
 class BaseVlmOptions(BaseModel):
     kind: str
-    prompt: str
+    prompt: Union[str, Callable[[Optional[SegmentedPage]], str]]
     scale: float = 2.0
     max_size: Optional[int] = None
+    temperature: float = 0.0
 
 
 class ResponseFormat(str, Enum):
@@ -29,6 +31,12 @@ class TransformersModelType(str, Enum):
     AUTOMODEL = "automodel"
     AUTOMODEL_VISION2SEQ = "automodel-vision2seq"
     AUTOMODEL_CAUSALLM = "automodel-causallm"
+    AUTOMODEL_IMAGETEXTTOTEXT = "automodel-imagetexttotext"
+
+
+class TransformersPromptStyle(str, Enum):
+    CHAT = "chat"
+    RAW = "raw"
 
 
 class InlineVlmOptions(BaseVlmOptions):
@@ -42,6 +50,7 @@ class InlineVlmOptions(BaseVlmOptions):
 
     inference_framework: InferenceFramework
     transformers_model_type: TransformersModelType = TransformersModelType.AUTOMODEL
+    transformers_prompt_style: TransformersPromptStyle = TransformersPromptStyle.CHAT
     response_format: ResponseFormat
 
     torch_dtype: Optional[str] = None
@@ -51,7 +60,6 @@ class InlineVlmOptions(BaseVlmOptions):
         AcceleratorDevice.MPS,
     ]
 
-    temperature: float = 0.0
     stop_strings: List[str] = []
     extra_generation_config: Dict[str, Any] = {}
 

{docling-2.40.0 → docling-2.42.0}/docling/document_converter.py

@@ -1,6 +1,7 @@
 import hashlib
 import logging
 import sys
+import threading
 import time
 from collections.abc import Iterable, Iterator
 from functools import partial
@@ -49,6 +50,7 @@ from docling.pipeline.standard_pdf_pipeline import StandardPdfPipeline
 from docling.utils.utils import chunkify
 
 _log = logging.getLogger(__name__)
+_PIPELINE_CACHE_LOCK = threading.Lock()
 
 
 class FormatOption(BaseModel):
@@ -315,17 +317,18 @@ class DocumentConverter:
         # Use a composite key to cache pipelines
         cache_key = (pipeline_class, options_hash)
 
-        if cache_key not in self.initialized_pipelines:
-            _log.info(
-                f"Initializing pipeline for {pipeline_class.__name__} with options hash {options_hash}"
-            )
-            self.initialized_pipelines[cache_key] = pipeline_class(
-                pipeline_options=pipeline_options
-            )
-        else:
-            _log.debug(
-                f"Reusing cached pipeline for {pipeline_class.__name__} with options hash {options_hash}"
-            )
+        with _PIPELINE_CACHE_LOCK:
+            if cache_key not in self.initialized_pipelines:
+                _log.info(
+                    f"Initializing pipeline for {pipeline_class.__name__} with options hash {options_hash}"
+                )
+                self.initialized_pipelines[cache_key] = pipeline_class(
+                    pipeline_options=pipeline_options
+                )
+            else:
+                _log.debug(
+                    f"Reusing cached pipeline for {pipeline_class.__name__} with options hash {options_hash}"
+                )
 
         return self.initialized_pipelines[cache_key]
 

{docling-2.40.0 → docling-2.42.0}/docling/models/api_vlm_model.py

@@ -29,12 +29,9 @@ class ApiVlmModel(BasePageModel):
 
             self.timeout = self.vlm_options.timeout
             self.concurrency = self.vlm_options.concurrency
-            self.prompt_content = (
-                f"This is a page from a document.\n{self.vlm_options.prompt}"
-            )
             self.params = {
                 **self.vlm_options.params,
-                "temperature": 0,
+                "temperature": self.vlm_options.temperature,
             }
 
     def __call__(
@@ -56,9 +53,14 @@ class ApiVlmModel(BasePageModel):
                         if hi_res_image.mode != "RGB":
                             hi_res_image = hi_res_image.convert("RGB")
 
+                    if callable(self.vlm_options.prompt):
+                        prompt = self.vlm_options.prompt(page.parsed_page)
+                    else:
+                        prompt = self.vlm_options.prompt
+
                     page_tags = api_image_request(
                         image=hi_res_image,
-                        prompt=self.prompt_content,
+                        prompt=prompt,
                         url=self.vlm_options.url,
                         timeout=self.timeout,
                         headers=self.vlm_options.headers,

{docling-2.40.0 → docling-2.42.0}/docling/models/document_picture_classifier.py

@@ -14,7 +14,8 @@ from PIL import Image
 from pydantic import BaseModel
 
 from docling.datamodel.accelerator_options import AcceleratorOptions
-from docling.models.base_model import BaseEnrichmentModel
+from docling.datamodel.base_models import ItemAndImageEnrichmentElement
+from docling.models.base_model import BaseItemAndImageEnrichmentModel
 from docling.models.utils.hf_model_download import download_hf_model
 from docling.utils.accelerator_utils import decide_device
 
@@ -32,7 +33,7 @@ class DocumentPictureClassifierOptions(BaseModel):
     kind: Literal["document_picture_classifier"] = "document_picture_classifier"
 
 
-class DocumentPictureClassifier(BaseEnrichmentModel):
+class DocumentPictureClassifier(BaseItemAndImageEnrichmentModel):
     """
     A model for classifying pictures in documents.
 
@@ -135,7 +136,7 @@ class DocumentPictureClassifier(BaseEnrichmentModel):
     def __call__(
         self,
         doc: DoclingDocument,
-        element_batch: Iterable[NodeItem],
+        element_batch: Iterable[ItemAndImageEnrichmentElement],
     ) -> Iterable[NodeItem]:
         """
         Processes a batch of elements and enriches them with classification predictions.
@@ -144,7 +145,7 @@ class DocumentPictureClassifier(BaseEnrichmentModel):
         ----------
         doc : DoclingDocument
             The document containing the elements to be processed.
-        element_batch : Iterable[NodeItem]
+        element_batch : Iterable[ItemAndImageEnrichmentElement]
             A batch of pictures to classify.
 
         Returns
@@ -155,22 +156,20 @@ class DocumentPictureClassifier(BaseEnrichmentModel):
         """
         if not self.enabled:
             for element in element_batch:
-                yield element
+                yield element.item
             return
 
         images: List[Union[Image.Image, np.ndarray]] = []
         elements: List[PictureItem] = []
         for el in element_batch:
-            assert isinstance(el, PictureItem)
-            elements.append(el)
-            img = el.get_image(doc)
-            assert img is not None
-            images.append(img)
+            assert isinstance(el.item, PictureItem)
+            elements.append(el.item)
+            images.append(el.image)
 
         outputs = self.document_picture_classifier.predict(images)
 
-        for element, output in zip(elements, outputs):
-            element.annotations.append(
+        for item, output in zip(elements, outputs):
+            item.annotations.append(
                 PictureClassificationData(
                     provenance="DocumentPictureClassifier",
                     predicted_classes=[
@@ -183,4 +182,4 @@ class DocumentPictureClassifier(BaseEnrichmentModel):
                 )
             )
 
-            yield element
+            yield item

{docling-2.40.0 → docling-2.42.0}/docling/models/layout_model.py

@@ -12,6 +12,7 @@ from PIL import Image
 from docling.datamodel.accelerator_options import AcceleratorOptions
 from docling.datamodel.base_models import BoundingBox, Cluster, LayoutPrediction, Page
 from docling.datamodel.document import ConversionResult
+from docling.datamodel.layout_model_specs import DOCLING_LAYOUT_V2, LayoutModelConfig
 from docling.datamodel.pipeline_options import LayoutOptions
 from docling.datamodel.settings import settings
 from docling.models.base_model import BasePageModel
@@ -25,9 +26,6 @@ _log = logging.getLogger(__name__)
 
 
 class LayoutModel(BasePageModel):
-    _model_repo_folder = "ds4sd--docling-models"
-    _model_path = "model_artifacts/layout"
-
     TEXT_ELEM_LABELS = [
         DocItemLabel.TEXT,
         DocItemLabel.FOOTNOTE,
@@ -59,25 +57,28 @@ class LayoutModel(BasePageModel):
         self.options = options
 
         device = decide_device(accelerator_options.device)
+        layout_model_config = options.model_spec
+        model_repo_folder = layout_model_config.model_repo_folder
+        model_path = layout_model_config.model_path
 
         if artifacts_path is None:
-            artifacts_path = self.download_models() / self._model_path
+            artifacts_path = (
+                self.download_models(layout_model_config=layout_model_config)
+                / model_path
+            )
         else:
-            # will become the default in the future
-            if (artifacts_path / self._model_repo_folder).exists():
-                artifacts_path = (
-                    artifacts_path / self._model_repo_folder / self._model_path
-                )
-            elif (artifacts_path / self._model_path).exists():
+            if (artifacts_path / model_repo_folder).exists():
+                artifacts_path = artifacts_path / model_repo_folder / model_path
+            elif (artifacts_path / model_path).exists():
                 warnings.warn(
                     "The usage of artifacts_path containing directly "
-                    f"{self._model_path} is deprecated. Please point "
+                    f"{model_path} is deprecated. Please point "
                     "the artifacts_path to the parent containing "
-                    f"the {self._model_repo_folder} folder.",
+                    f"the {model_repo_folder} folder.",
                     DeprecationWarning,
                     stacklevel=3,
                 )
-                artifacts_path = artifacts_path / self._model_path
+                artifacts_path = artifacts_path / model_path
 
         self.layout_predictor = LayoutPredictor(
             artifact_path=str(artifacts_path),
@@ -90,10 +91,11 @@ class LayoutModel(BasePageModel):
         local_dir: Optional[Path] = None,
         force: bool = False,
         progress: bool = False,
+        layout_model_config: LayoutModelConfig = DOCLING_LAYOUT_V2,
     ) -> Path:
         return download_hf_model(
-            repo_id="ds4sd/docling-models",
-            revision="v2.2.0",
+            repo_id=layout_model_config.repo_id,
+            revision=layout_model_config.revision,
             local_dir=local_dir,
             force=force,
             progress=progress,