numind 0.4.1__tar.gz → 0.4.2__tar.gz

{numind-0.4.1 → numind-0.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: numind
-Version: 0.4.1
+Version: 0.4.2
 Summary: SDK to interact with the NuMind models API.
 Project-URL: Homepage, https://github.com/numindai/nuextract-platform-sdk
 Project-URL: Repository, https://github.com/numindai/nuextract-platform-sdk.git
@@ -441,6 +441,7 @@ Class | Method | HTTP request | Description
  - [ContentExtractionResponse](docs/ContentExtractionResponse.md)
  - [ContentProjectResponse](docs/ContentProjectResponse.md)
  - [ContentProjectSettingsResponse](docs/ContentProjectSettingsResponse.md)
+ - [ContentProjectSettingsResponseImageWidth](docs/ContentProjectSettingsResponseImageWidth.md)
  - [ContentProjectSettingsResponseRasterizationDPI](docs/ContentProjectSettingsResponseRasterizationDPI.md)
  - [ContentSettingsPatch](docs/ContentSettingsPatch.md)
  - [ConvertRequest](docs/ConvertRequest.md)

{numind-0.4.1 → numind-0.4.2}/README.md

@@ -365,6 +365,7 @@ Class | Method | HTTP request | Description
  - [ContentExtractionResponse](docs/ContentExtractionResponse.md)
  - [ContentProjectResponse](docs/ContentProjectResponse.md)
  - [ContentProjectSettingsResponse](docs/ContentProjectSettingsResponse.md)
+ - [ContentProjectSettingsResponseImageWidth](docs/ContentProjectSettingsResponseImageWidth.md)
  - [ContentProjectSettingsResponseRasterizationDPI](docs/ContentProjectSettingsResponseRasterizationDPI.md)
  - [ContentSettingsPatch](docs/ContentSettingsPatch.md)
  - [ConvertRequest](docs/ConvertRequest.md)

{numind-0.4.1 → numind-0.4.2}/docs/ContentExtractionApi.md

@@ -16,6 +16,7 @@ Method | HTTP request | Description
 
 #### Response:
  Returns a markdown string representing the extracted content.
+ `isTruncated` is true when at least one document part stops at a token limit; the combined result may be incomplete.
 
 #### Error Responses:
 `404 Not Found` - If a content extraction job with the specified ID does not exist.
@@ -101,7 +102,7 @@ Name | Type | Description  | Notes
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
 # **post_api_content_extraction_jobs**
-> JobIdResponse post_api_content_extraction_jobs(file, x_organization_id=x_organization_id, project_id=project_id, temperature=temperature, rasterization_dpi=rasterization_dpi, max_output_tokens=max_output_tokens, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, timeout=timeout)
+> JobIdResponse post_api_content_extraction_jobs(file, x_organization_id=x_organization_id, project_id=project_id, temperature=temperature, rasterization_dpi=rasterization_dpi, max_output_tokens=max_output_tokens, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, image_width=image_width, timeout=timeout)
 
 
  Extract content from the provided file in markdown format as an async job.
@@ -121,7 +122,7 @@ Name | Type | Description  | Notes
 #### Settings:
 Request settings override the project settings when `projectId` is provided; otherwise omitted settings use platform defaults.
 Platform defaults: temperature 0.8, rasterization DPI `automatic`,
-output token budget 10000, thinking enabled, thinking token budget 15000,
+output token budget 20000, thinking enabled, thinking token budget 15000,
 loop detection enabled, and random seed `2022`.
 
 
@@ -155,13 +156,14 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     project_id = "project_id_example"  # str | Optional content project identifier. When provided, project settings are used as fallback for inference parameters; this identifier is also attached to usage and billing events. (optional)
     temperature = 3.4  # float | Model temperature. Controls output diversity. Ranges between 0 and 2. (optional)
     rasterization_dpi = "rasterization_dpi_example"  # str | Resolution used to convert formatted documents (PDFs, etc.) to images, in dots per inch. Use 'automatic' or an integer between 1 and 600. Numeric strings are also accepted. Ignored for files that are already images or text. (optional)
-    max_output_tokens = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    max_output_tokens = "max_output_tokens_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     enable_thinking = True  # bool | Enable thinking/reasoning. (optional)
-    thinking_token_budget = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    thinking_token_budget = "thinking_token_budget_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     loop_detection = (
         True  # bool | Stop the generation when the model repeats itself. (optional)
     )
     random_seed = "random_seed_example"  # str | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. (optional)
+    image_width = "image_width_example"  # str | Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. (optional)
     timeout = "timeout_example"  # str | Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h (optional)
 
     try:
@@ -176,6 +178,7 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
             thinking_token_budget=thinking_token_budget,
             loop_detection=loop_detection,
             random_seed=random_seed,
+            image_width=image_width,
             timeout=timeout,
         )
         print(
@@ -201,11 +204,12 @@ Name | Type | Description  | Notes
  **project_id** | **str**| Optional content project identifier. When provided, project settings are used as fallback for inference parameters; this identifier is also attached to usage and billing events. | [optional] 
  **temperature** | **float**| Model temperature. Controls output diversity. Ranges between 0 and 2. | [optional] 
  **rasterization_dpi** | **str**| Resolution used to convert formatted documents (PDFs, etc.) to images, in dots per inch. Use 'automatic' or an integer between 1 and 600. Numeric strings are also accepted. Ignored for files that are already images or text. | [optional] 
- **max_output_tokens** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **max_output_tokens** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **enable_thinking** | **bool**| Enable thinking/reasoning. | [optional] 
- **thinking_token_budget** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **thinking_token_budget** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **loop_detection** | **bool**| Stop the generation when the model repeats itself. | [optional] 
  **random_seed** | **str**| Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | [optional] 
+ **image_width** | **str**| Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. | [optional] 
  **timeout** | **str**| Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h | [optional] 
 
 ### Return type
@@ -226,7 +230,7 @@ Name | Type | Description  | Notes
 | Status code | Description | Response headers |
 |-------------|-------------|------------------|
 **200** |  |  -  |
-**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxOutputTokens, Invalid value for: query parameter enableThinking, Invalid value for: query parameter thinkingTokenBudget, Invalid value for: query parameter loopDetection, Invalid value for: body |  -  |
+**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter enableThinking, Invalid value for: query parameter loopDetection, Invalid value for: body |  -  |
 **0** |  |  -  |
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)

{numind-0.4.1 → numind-0.4.2}/docs/ContentExtractionProjectManagementApi.md

@@ -315,11 +315,12 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     content_settings_patch = {
         "temperature": 0.8,
         "rasterizationDPI": "automatic",
-        "maxOutputTokens": 10000,
+        "maxOutputTokens": "20000",
         "enableThinking": true,
-        "thinkingTokenBudget": 15000,
+        "thinkingTokenBudget": "15000",
         "loopDetection": true,
         "randomSeed": "2022",
+        "imageWidth": "automatic",
     }  # ContentSettingsPatch |
     x_organization_id = "x_organization_id_example"  # str | Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. (optional)
 
@@ -396,9 +397,10 @@ A **Project** is created with default settings:
 ---------|---------|
  `temperature` | 0.8 |
  `rasterizationDPI` | `automatic` |
- `maxOutputTokens` | 10000 |
+ `imageWidth` | `automatic` |
+ `maxOutputTokens` | 20000 |
  `enableThinking` | true |
- `thinkingTokenBudget` | 15000 (0 for no limit) |
+ `thinkingTokenBudget` | 15000 (`Unlimited` for no limit) |
  `loopDetection` | true |
  `randomSeed` | 2022 (-9223372036854775808 to 9223372036854775807, or `random`) |
 
@@ -503,9 +505,9 @@ Default values are:
 -----------|--------|
  `temperature` | 0.8 |
  `rasterizationDPI` | `automatic` |
- `maxOutputTokens` | 10000 |
+ `maxOutputTokens` | 20000 |
  `enableThinking` | true |
- `thinkingTokenBudget` | 15000 (0 for no limit) |
+ `thinkingTokenBudget` | 15000 (`Unlimited` for no limit) |
  `loopDetection` | true |
  `randomSeed` | 2022 (-9223372036854775808 to 9223372036854775807, or `random`) |
 

{numind-0.4.1 → numind-0.4.2}/docs/ContentExtractionResponse.md

@@ -14,6 +14,7 @@ Name | Type | Description | Notes
 **logprobs** | **float** | Logprob of the inference result (sum of logprobs of all tokens). | 
 **output_token_probability** | **float** | Geometric mean of the output token probabilities. | 
 **random_seed** | **int** | Random seed used for this inference. | 
+**is_truncated** | **bool** | True when at least one document part stopped at a token limit. The result may be incomplete. | 
 
 ## Example
 

{numind-0.4.1 → numind-0.4.2}/docs/ContentProjectSettingsResponse.md

@@ -8,11 +8,12 @@ Name | Type | Description | Notes
 ------------ | ------------- | ------------- | -------------
 **temperature** | **float** | Model temperature. Controls output diversity. Ranges between 0 and 2. | 
 **rasterization_dpi** | [**ContentProjectSettingsResponseRasterizationDPI**](ContentProjectSettingsResponseRasterizationDPI.md) |  | 
-**max_output_tokens** | **int** | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | 
+**max_output_tokens** | **str** | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | 
 **enable_thinking** | **bool** | Enable thinking/reasoning. | 
-**thinking_token_budget** | **int** | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | 
+**thinking_token_budget** | **str** | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | 
 **loop_detection** | **bool** | Stop the generation when the model repeats itself. | 
 **random_seed** | **str** | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | 
+**image_width** | [**ContentProjectSettingsResponseImageWidth**](ContentProjectSettingsResponseImageWidth.md) |  | 
 
 ## Example
 

numind-0.4.2/docs/ContentProjectSettingsResponseImageWidth.md

@@ -0,0 +1,39 @@
+# ContentProjectSettingsResponseImageWidth
+
+Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed.
+
+## Properties
+
+Name | Type | Description | Notes
+------------ | ------------- | ------------- | -------------
+
+## Example
+
+```python
+from numind.models.content_project_settings_response_image_width import (
+    ContentProjectSettingsResponseImageWidth,
+)
+
+# TODO update the JSON string below
+json = "{}"
+# create an instance of ContentProjectSettingsResponseImageWidth from a JSON string
+content_project_settings_response_image_width_instance = (
+    ContentProjectSettingsResponseImageWidth.from_json(json)
+)
+# print the JSON string representation of the object
+print(ContentProjectSettingsResponseImageWidth.to_json())
+
+# convert the object into a dict
+content_project_settings_response_image_width_dict = (
+    content_project_settings_response_image_width_instance.to_dict()
+)
+# create an instance of ContentProjectSettingsResponseImageWidth from a dict
+content_project_settings_response_image_width_from_dict = (
+    ContentProjectSettingsResponseImageWidth.from_dict(
+        content_project_settings_response_image_width_dict
+    )
+)
+```
+[[Back to Model list]](../README.md#documentation-for-models) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to README]](../README.md)
+
+

{numind-0.4.1 → numind-0.4.2}/docs/ContentSettingsPatch.md

@@ -7,11 +7,12 @@ Name | Type | Description | Notes
 ------------ | ------------- | ------------- | -------------
 **temperature** | **float** | Model temperature. Controls output diversity. Ranges between 0 and 2. | [optional] 
 **rasterization_dpi** | [**ContentProjectSettingsResponseRasterizationDPI**](ContentProjectSettingsResponseRasterizationDPI.md) |  | [optional] 
-**max_output_tokens** | **int** | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+**max_output_tokens** | **str** | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
 **enable_thinking** | **bool** | Enable thinking/reasoning. | [optional] 
-**thinking_token_budget** | **int** | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+**thinking_token_budget** | **str** | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
 **loop_detection** | **bool** | Stop the generation when the model repeats itself. | [optional] 
 **random_seed** | **str** | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | [optional] 
+**image_width** | [**ContentProjectSettingsResponseImageWidth**](ContentProjectSettingsResponseImageWidth.md) |  | [optional] 
 
 ## Example
 

{numind-0.4.1 → numind-0.4.2}/docs/ConvertRequest.md

@@ -7,6 +7,7 @@ Name | Type | Description | Notes
 ------------ | ------------- | ------------- | -------------
 **rasterization_dpi** | [**ContentProjectSettingsResponseRasterizationDPI**](ContentProjectSettingsResponseRasterizationDPI.md) |  | 
 **token_budget_type** | **str** | Token budget allocation used while converting the document: `per-page` applies a separate budget to each page, while `all-pages` shares one budget across the entire document. | 
+**image_width** | [**ContentProjectSettingsResponseImageWidth**](ContentProjectSettingsResponseImageWidth.md) |  | 
 
 ## Example
 

{numind-0.4.1 → numind-0.4.2}/docs/FilesApi.md

@@ -260,7 +260,12 @@ Name | Type | Description  | Notes
 
  Convert the **File** into a **Document**
 
- - For ***text and image files***, the content is used as-is — no conversion is performed.
+ - For ***text files***, the content is used as-is.
+ - For ***original standalone raster images***, optional `imageWidth` controls resizing:
+ `"automatic"` sets the short side to 1280 pixels, a positive integer sets horizontal width,
+ and `"original"` bypasses processing. Omission defaults to `"original"`.
+ Aspect ratio is preserved and upscaling is allowed. Original upload bytes are retained.
+ This setting never applies to images produced from PDF, TIFF, or other document conversions.
  - For ***other supported file types*** (e.g., PDFs, WORD, PPTX, Excel),
  the file is ***converted to an image*** in the background,
  using the **conversion parameters** provided in the request body (e.g., `rasterizationDPI`).
@@ -269,7 +274,7 @@ Example request bodies: `{"rasterizationDPI":"automatic","tokenBudgetType":"all-
 `{"rasterizationDPI":155,"tokenBudgetType":"per-page"}` for fixed resolution, or `{"rasterizationDPI":"200","tokenBudgetType":"all-pages"}` for a numeric string.
 The `automatic` keyword is case-insensitive. Fixed DPI must be a whole number from 1 to 600, inclusive;
 responses encode fixed DPI as a number. Unlike a partial project settings update, this body requires
-`rasterizationDPI` and `tokenBudgetType` (`all-pages` or `per-page`): an omitted or `null` value is rejected,
+`rasterizationDPI` and `tokenBudgetType` (`all-pages` or `per-page`) are required; omitted or `null` values are rejected,
 as are `0`, `601`, and `155.5` for `rasterizationDPI`.
 
  The resulting image is then saved as a **Document** and can be used for inference or further processing.

{numind-0.4.1 → numind-0.4.2}/docs/InferenceApi.md

@@ -121,7 +121,7 @@ Name | Type | Description  | Notes
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
 # **post_api_structured_extraction_structuredprojectid_jobs_document_documentid**
-> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs_document_documentid(structured_project_id, document_id, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, timeout=timeout)
+> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs_document_documentid(structured_project_id, document_id, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, image_width=image_width, timeout=timeout)
 
 
  Performs structured information extraction inference on the provided document as an async job.
@@ -179,16 +179,17 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     document_id = "document_id_example"  # str | Unique document identifier.
     x_organization_id = "x_organization_id_example"  # str | Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. (optional)
     temperature = 3.4  # float | Model temperature. Controls output diversity. Ranges between 0 and 2. (optional)
-    max_output_tokens = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    max_output_tokens = "max_output_tokens_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     max_example_token_number = 56  # int | Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. (optional)
-    max_example_number = 56  # int | Maximum number of examples to use. Must be positive. Set to 0 for no limit. (optional)
+    max_example_number = "max_example_number_example"  # str | Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. (optional)
     min_example_similarity = 3.4  # float | Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. (optional)
     enable_thinking = True  # bool | Enable thinking/reasoning. (optional)
-    thinking_token_budget = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    thinking_token_budget = "thinking_token_budget_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     loop_detection = (
         True  # bool | Stop the generation when the model repeats itself. (optional)
     )
     random_seed = "random_seed_example"  # str | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. (optional)
+    image_width = "image_width_example"  # str | Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. (optional)
     timeout = "timeout_example"  # str | Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h (optional)
 
     try:
@@ -205,6 +206,7 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
             thinking_token_budget=thinking_token_budget,
             loop_detection=loop_detection,
             random_seed=random_seed,
+            image_width=image_width,
             timeout=timeout,
         )
         print(
@@ -229,14 +231,15 @@ Name | Type | Description  | Notes
  **document_id** | **str**| Unique document identifier. | 
  **x_organization_id** | **str**| Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. | [optional] 
  **temperature** | **float**| Model temperature. Controls output diversity. Ranges between 0 and 2. | [optional] 
- **max_output_tokens** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **max_output_tokens** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **max_example_token_number** | **int**| Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. | [optional] 
- **max_example_number** | **int**| Maximum number of examples to use. Must be positive. Set to 0 for no limit. | [optional] 
+ **max_example_number** | **str**| Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. | [optional] 
  **min_example_similarity** | **float**| Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. | [optional] 
  **enable_thinking** | **bool**| Enable thinking/reasoning. | [optional] 
- **thinking_token_budget** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **thinking_token_budget** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **loop_detection** | **bool**| Stop the generation when the model repeats itself. | [optional] 
  **random_seed** | **str**| Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | [optional] 
+ **image_width** | **str**| Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. | [optional] 
  **timeout** | **str**| Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h | [optional] 
 
 ### Return type
@@ -257,13 +260,13 @@ Name | Type | Description  | Notes
 | Status code | Description | Response headers |
 |-------------|-------------|------------------|
 **200** |  |  -  |
-**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxOutputTokens, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter maxExampleNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter thinkingTokenBudget, Invalid value for: query parameter loopDetection |  -  |
+**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter loopDetection |  -  |
 **0** |  |  -  |
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
 # **post_api_template_generation_jobs_document_documentid**
-> JobIdResponse post_api_template_generation_jobs_document_documentid(document_id, x_organization_id=x_organization_id, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, timeout=timeout)
+> JobIdResponse post_api_template_generation_jobs_document_documentid(document_id, x_organization_id=x_organization_id, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, image_width=image_width, timeout=timeout)
 
 
  Derive a template from the provided **Document** as an async job.
@@ -314,10 +317,11 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     document_id = "document_id_example"  # str | Unique document identifier.
     x_organization_id = "x_organization_id_example"  # str | Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. (optional)
     enable_thinking = True  # bool | Enable thinking/reasoning. (optional)
-    thinking_token_budget = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    thinking_token_budget = "thinking_token_budget_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     loop_detection = (
         True  # bool | Stop the generation when the model repeats itself. (optional)
     )
+    image_width = "image_width_example"  # str | Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. (optional)
     timeout = "timeout_example"  # str | Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h (optional)
 
     try:
@@ -328,6 +332,7 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
                 enable_thinking=enable_thinking,
                 thinking_token_budget=thinking_token_budget,
                 loop_detection=loop_detection,
+                image_width=image_width,
                 timeout=timeout,
             )
         )
@@ -352,8 +357,9 @@ Name | Type | Description  | Notes
  **document_id** | **str**| Unique document identifier. | 
  **x_organization_id** | **str**| Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. | [optional] 
  **enable_thinking** | **bool**| Enable thinking/reasoning. | [optional] 
- **thinking_token_budget** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **thinking_token_budget** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **loop_detection** | **bool**| Stop the generation when the model repeats itself. | [optional] 
+ **image_width** | **str**| Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. | [optional] 
  **timeout** | **str**| Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h | [optional] 
 
 ### Return type
@@ -374,7 +380,7 @@ Name | Type | Description  | Notes
 | Status code | Description | Response headers |
 |-------------|-------------|------------------|
 **200** |  |  -  |
-**400** | Invalid value for: query parameter enableThinking, Invalid value for: query parameter thinkingTokenBudget, Invalid value for: query parameter loopDetection |  -  |
+**400** | Invalid value for: query parameter enableThinking, Invalid value for: query parameter loopDetection |  -  |
 **0** |  |  -  |
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)

{numind-0.4.1 → numind-0.4.2}/docs/StructuredDataExtractionApi.md

@@ -20,7 +20,8 @@ Method | HTTP request | Description
  Returns a JSON representing the extracted information.
  In case the extraction fails, an empty template is returned.
  HTTP code 206 is returned when the response contains an inference error, including whenever the model does not finish normally.
- Reaching the output token limit is reported as truncation.
+ `isTruncated` is true when the model stops at a token limit, even if the output also contains invalid JSON.
+ The existing `error` field and HTTP status still report validation errors; `isTruncated` independently indicates that the result may be incomplete.
 
 #### Error Responses:
 `404 Not Found` - If an extraction job with the specified ID does not exist.
@@ -120,6 +121,8 @@ Name | Type | Description  | Notes
 
  The `request` JSON contains the schema, optional instructions, optional settings, and optional example outputs.
  Example files themselves are not embedded in JSON; they are attached as `exampleFiles` multipart parts.
+ `settings.imageWidth` accepts `"automatic"`, `"original"`, or a positive integer width.
+ It applies only to original standalone images, including uploaded examples; converted document pages use DPI.
  `settings.rasterizationDPI` accepts `"automatic"` or an integer from 1 to 600, such as `200`. Numeric strings such as `"200"` are also accepted.
 
 #### Request JSON:
@@ -135,12 +138,12 @@ Name | Type | Description  | Notes
   "settings": {
     "temperature": 0.2,
     "rasterizationDPI": 200,
-    "maxOutputTokens": 2048,
+    "maxOutputTokens": "2048",
     "maxExampleTokenNumber": 12000,
-    "maxExampleNumber": 3,
+    "maxExampleNumber": "3",
     "minExampleSimilarity": 0,
     "enableThinking": false,
-    "thinkingTokenBudget": 8000,
+    "thinkingTokenBudget": "8000",
     "loopDetection": true,
     "randomSeed": "99"
   },
@@ -179,6 +182,7 @@ curl -X POST "$BASE_URL/structured-extraction/jobs" \
 `maxExampleTokenNumber`, `maxExampleNumber`, `minExampleSimilarity`, `enableThinking`,
 `thinkingTokenBudget`, `loopDetection` and `randomSeed`. `randomSeed` must be a string containing a 64-bit integer or `random`.
 Omitted settings use the server defaults for structured extraction.
+`maxExampleNumber` defaults to `"5"`; use `"0"` for no examples, a positive integer string for a limit, or `"Unlimited"` for no count limit.
 
 `examples` is optional. Each example must contain a `result` object that conforms to `schema`.
 
@@ -277,7 +281,7 @@ Name | Type | Description  | Notes
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
 # **post_api_structured_extraction_structuredprojectid_jobs**
-> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs(structured_project_id, body, x_organization_id=x_organization_id, dpi=dpi, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, timeout=timeout)
+> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs(structured_project_id, body, x_organization_id=x_organization_id, dpi=dpi, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, image_width=image_width, timeout=timeout)
 
 
  Extract structured information from the provided text or file as an async job. Some files are converted to images -
@@ -336,16 +340,17 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     x_organization_id = "x_organization_id_example"  # str | Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. (optional)
     dpi = "dpi_example"  # str | Resolution used to convert formatted documents (PDFs, etc.) to images, in dots per inch. Use 'automatic' or an integer between 1 and 600. Numeric strings are also accepted. Ignored for files that are already images or text. (optional)
     temperature = 3.4  # float | Model temperature. Controls output diversity. Ranges between 0 and 2. (optional)
-    max_output_tokens = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    max_output_tokens = "max_output_tokens_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     max_example_token_number = 56  # int | Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. (optional)
-    max_example_number = 56  # int | Maximum number of examples to use. Must be positive. Set to 0 for no limit. (optional)
+    max_example_number = "max_example_number_example"  # str | Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. (optional)
     min_example_similarity = 3.4  # float | Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. (optional)
     enable_thinking = True  # bool | Enable thinking/reasoning. (optional)
-    thinking_token_budget = 56  # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    thinking_token_budget = "thinking_token_budget_example"  # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     loop_detection = (
         True  # bool | Stop the generation when the model repeats itself. (optional)
     )
     random_seed = "random_seed_example"  # str | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. (optional)
+    image_width = "image_width_example"  # str | Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. (optional)
     timeout = "timeout_example"  # str | Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h (optional)
 
     try:
@@ -364,6 +369,7 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
                 thinking_token_budget=thinking_token_budget,
                 loop_detection=loop_detection,
                 random_seed=random_seed,
+                image_width=image_width,
                 timeout=timeout,
             )
         )
@@ -390,14 +396,15 @@ Name | Type | Description  | Notes
  **x_organization_id** | **str**| Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. | [optional] 
  **dpi** | **str**| Resolution used to convert formatted documents (PDFs, etc.) to images, in dots per inch. Use 'automatic' or an integer between 1 and 600. Numeric strings are also accepted. Ignored for files that are already images or text. | [optional] 
  **temperature** | **float**| Model temperature. Controls output diversity. Ranges between 0 and 2. | [optional] 
- **max_output_tokens** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **max_output_tokens** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **max_example_token_number** | **int**| Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. | [optional] 
- **max_example_number** | **int**| Maximum number of examples to use. Must be positive. Set to 0 for no limit. | [optional] 
+ **max_example_number** | **str**| Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. | [optional] 
  **min_example_similarity** | **float**| Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. | [optional] 
  **enable_thinking** | **bool**| Enable thinking/reasoning. | [optional] 
- **thinking_token_budget** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **thinking_token_budget** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **loop_detection** | **bool**| Stop the generation when the model repeats itself. | [optional] 
  **random_seed** | **str**| Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | [optional] 
+ **image_width** | **str**| Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. | [optional] 
  **timeout** | **str**| Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h | [optional] 
 
 ### Return type
@@ -418,13 +425,13 @@ Name | Type | Description  | Notes
 | Status code | Description | Response headers |
 |-------------|-------------|------------------|
 **200** |  |  -  |
-**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxOutputTokens, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter maxExampleNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter thinkingTokenBudget, Invalid value for: query parameter loopDetection, Invalid value |  -  |
+**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter loopDetection, Invalid value |  -  |
 **0** |  |  -  |
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)
 
 # **post_api_structured_extraction_structuredprojectid_jobs_text**
-> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs_text(structured_project_id, text_request, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, timeout=timeout)
+> JobIdResponse post_api_structured_extraction_structuredprojectid_jobs_text(structured_project_id, text_request, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, image_width=image_width, timeout=timeout)
 
 
  Performs structured information extraction inference on the provided text as an async job.
@@ -486,18 +493,19 @@ with numind.openapi_client.ApiClient(configuration) as api_client:
     text_request = {text=[EXAMPLE ONLY] Your order (ID: o-89123) has been successfully processed. The customer ID for this order is c-20485. It was placed on March 10, 2024, at 11:15 AM UTC and is now marked as shipped. The total amount charged is $149.99 USD. The items in this order include: Product ID p-00876 with a quantity of 1 at a unit price of $79.99, and Product ID p-00321 with a quantity of 2 at a unit price of $35.00. The shipping address is 782 Pine St, Austin, TX, 73301, USA. The customer has requested: "Leave package at the front door." Additional delivery preferences include no signature required and standard delivery. The estimated delivery date is March 15, 2024, by 5:00 PM UTC.} # TextRequest | 
     x_organization_id = 'x_organization_id_example' # str | Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. (optional)
     temperature = 3.4 # float | Model temperature. Controls output diversity. Ranges between 0 and 2. (optional)
-    max_output_tokens = 56 # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    max_output_tokens = 'max_output_tokens_example' # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     max_example_token_number = 56 # int | Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. (optional)
-    max_example_number = 56 # int | Maximum number of examples to use. Must be positive. Set to 0 for no limit. (optional)
+    max_example_number = 'max_example_number_example' # str | Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. (optional)
     min_example_similarity = 3.4 # float | Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. (optional)
     enable_thinking = True # bool | Enable thinking/reasoning. (optional)
-    thinking_token_budget = 56 # int | Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. (optional)
+    thinking_token_budget = 'thinking_token_budget_example' # str | Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. (optional)
     loop_detection = True # bool | Stop the generation when the model repeats itself. (optional)
     random_seed = 'random_seed_example' # str | Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. (optional)
+    image_width = 'image_width_example' # str | Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. (optional)
     timeout = 'timeout_example' # str | Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h (optional)
 
     try:
-        api_response = api_instance.post_api_structured_extraction_structuredprojectid_jobs_text(structured_project_id, text_request, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, timeout=timeout)
+        api_response = api_instance.post_api_structured_extraction_structuredprojectid_jobs_text(structured_project_id, text_request, x_organization_id=x_organization_id, temperature=temperature, max_output_tokens=max_output_tokens, max_example_token_number=max_example_token_number, max_example_number=max_example_number, min_example_similarity=min_example_similarity, enable_thinking=enable_thinking, thinking_token_budget=thinking_token_budget, loop_detection=loop_detection, random_seed=random_seed, image_width=image_width, timeout=timeout)
         print("The response of StructuredDataExtractionApi->post_api_structured_extraction_structuredprojectid_jobs_text:\n")
         pprint(api_response)
     except Exception as e:
@@ -515,14 +523,15 @@ Name | Type | Description  | Notes
  **text_request** | [**TextRequest**](TextRequest.md)|  | 
  **x_organization_id** | **str**| Optional organization to use for this request.   No header means that the user personal account will be used.   This token is *only* used by the _frontend_ application and *will be ignored if used with the API*. When using the api, the organization used will be the one of the api key. | [optional] 
  **temperature** | **float**| Model temperature. Controls output diversity. Ranges between 0 and 2. | [optional] 
- **max_output_tokens** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **max_output_tokens** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **max_example_token_number** | **int**| Maximum number of tokens allocated to smart examples. Must be between 0 and 130000, within the model's context window. | [optional] 
- **max_example_number** | **int**| Maximum number of examples to use. Must be positive. Set to 0 for no limit. | [optional] 
+ **max_example_number** | **str**| Maximum number of examples to use. Given as a string: 0 for no examples, a positive integer string for a limit, or Unlimited for no count limit. | [optional] 
  **min_example_similarity** | **float**| Minimum similarity between the document and the examples. Must be between 0 and 1. Set to 0 for any similarity and 1 for exact match. | [optional] 
  **enable_thinking** | **bool**| Enable thinking/reasoning. | [optional] 
- **thinking_token_budget** | **int**| Maximum number of tokens for the corresponding budget (output or thinking). Use a positive integer, or 0 for no limit. The thinking budget is only used when thinking is enabled. | [optional] 
+ **thinking_token_budget** | **str**| Maximum number of tokens for the corresponding budget (output or thinking). Given as a string: a strictly positive integer string, or Unlimited for no limit; zero is invalid. The thinking budget is only used when thinking is enabled. | [optional] 
  **loop_detection** | **bool**| Stop the generation when the model repeats itself. | [optional] 
  **random_seed** | **str**| Inference seed setting. Use a string containing a 64-bit integer for a fixed seed, or `random` to generate one for each inference. | [optional] 
+ **image_width** | **str**| Original standalone images only (not PDF/TIFF or converted pages). Use automatic for a 1280 px short side, original to preserve the upload, or a positive integer for the upright image width in pixels. Aspect ratio is preserved; upscaling is allowed. | [optional] 
  **timeout** | **str**| Execution timeout for the async inference job. If omitted, the server default of 60m is used.   If provided below the server-configured minimum (5m by default), the effective timeout is clamped to that minimum.   Format examples: 1000ms, 10s, 1m, 1h | [optional] 
 
 ### Return type
@@ -543,7 +552,7 @@ Name | Type | Description  | Notes
 | Status code | Description | Response headers |
 |-------------|-------------|------------------|
 **200** |  |  -  |
-**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxOutputTokens, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter maxExampleNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter thinkingTokenBudget, Invalid value for: query parameter loopDetection, Invalid value for: body |  -  |
+**400** | Invalid value for: query parameter temperature, Invalid value for: query parameter maxExampleTokenNumber, Invalid value for: query parameter minExampleSimilarity, Invalid value for: query parameter enableThinking, Invalid value for: query parameter loopDetection, Invalid value for: body |  -  |
 **0** |  |  -  |
 
 [[Back to top]](#) [[Back to API list]](../README.md#documentation-for-api-endpoints) [[Back to Model list]](../README.md#documentation-for-models) [[Back to README]](../README.md)