Vertex AI OCR

總覽

屬性	詳細資訊
說明	Vertex AI OCR 由 Mistral 驅動，提供文件智慧能力，可從 PDF 與圖片中擷取文字
LiteLLM 上的提供者路由	vertex_ai/
支援的操作	/ocr
提供者文件連結	Vertex AI ↗

使用由 Mistral 提供支援的 Vertex AI OCR 模型，從文件與圖片中擷取文字。

快速開始

LiteLLM SDK

SDK Usage

import litellm
import os

# Set environment variables
os.environ["VERTEXAI_PROJECT"] = "your-project-id"
os.environ["VERTEXAI_LOCATION"] = "us-central1"

# OCR with PDF URL
response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={
        "type": "document_url",
        "document_url": "https://example.com/document.pdf"
    }
)

# Access extracted text
for page in response.pages:
    print(page.text)

LiteLLM PROXY

proxy_config.yaml

model_list:
  - model_name: vertex-ocr
    litellm_params:
      model: vertex_ai/mistral-ocr-2505
      vertex_project: os.environ/VERTEXAI_PROJECT
      vertex_location: os.environ/VERTEXAI_LOCATION
      vertex_credentials: path/to/service-account.json  # Optional
    model_info:
      mode: ocr

啟動 Proxy

litellm --config proxy_config.yaml

透過 Proxy 呼叫 OCR

cURL Request

curl -X POST http://localhost:4000/ocr \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer your-api-key" \
  -d '{
    "model": "vertex-ocr",
    "document": {
      "type": "document_url",
      "document_url": "https://arxiv.org/pdf/2201.04234"
    }
  }'

驗證

Vertex AI OCR 支援多種驗證方式：

服務帳戶 JSON

Service Account Auth

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={"type": "document_url", "document_url": "https://..."},
    vertex_project="your-project-id",
    vertex_location="us-central1",
    vertex_credentials="path/to/service-account.json"
)

應用程式預設憑證

Default Credentials

# Relies on GOOGLE_APPLICATION_CREDENTIALS environment variable
response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={"type": "document_url", "document_url": "https://..."},
    vertex_project="your-project-id",
    vertex_location="us-central1"
)

文件類型

Vertex AI OCR 同時支援 PDF 與圖片。

PDF 文件

PDF OCR

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={
        "type": "document_url",
        "document_url": "https://example.com/document.pdf"
    },
    vertex_project="your-project-id",
    vertex_location="us-central1"
)

圖片文件

Image OCR

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={
        "type": "image_url",
        "image_url": "https://example.com/image.png"
    },
    vertex_project="your-project-id",
    vertex_location="us-central1"
)

Base64 編碼文件

Base64 PDF

import base64

# Read and encode PDF
with open("document.pdf", "rb") as f:
    pdf_base64 = base64.b64encode(f.read()).decode()

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505", # This doesn't work for deepseek
    document={
        "type": "document_url",
        "document_url": f"data:application/pdf;base64,{pdf_base64}"
    },
    vertex_project="your-project-id",
    vertex_location="us-central1"
)

支援的參數

All Parameters

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={                           # Required: Document to process
        "type": "document_url",
        "document_url": "https://..."
    },
    vertex_project="your-project-id",   # Required: GCP project ID
    vertex_location="us-central1",       # Optional: Defaults to us-central1
    vertex_credentials="path/to/key.json", # Optional: Service account key
    include_image_base64=True,           # Optional: Include base64 images
    pages=[0, 1, 2],                     # Optional: Specific pages to process
    image_limit=10                       # Optional: Limit number of images
)

回應格式

Response Structure

# Response has the following structure
response.pages          # List of pages with extracted text
response.model          # Model used
response.object         # "ocr"
response.usage_info     # Token usage information

# Access page content
for page in response.pages:
    print(f"Page {page.page_number}:")
    print(page.text)

非同步支援

Async Usage

import litellm

response = await litellm.aocr(
    model="vertex_ai/mistral-ocr-2505",
    document={
        "type": "document_url",
        "document_url": "https://example.com/document.pdf"
    },
    vertex_project="your-project-id",
    vertex_location="us-central1"
)

成本追蹤

LiteLLM 會自動追蹤 Vertex AI OCR 的成本：

• 每頁成本：$0.0005（以每 1,000 頁 $1.50 計算）

View Cost

response = litellm.ocr(
    model="vertex_ai/mistral-ocr-2505",
    document={"type": "document_url", "document_url": "https://..."},
    vertex_project="your-project-id"
)

# Access cost information
print(f"Cost: ${response._hidden_params.get('response_cost', 0)}")

重要注意事項

URL 轉換

Vertex AI Mistral OCR 端點沒有網際網路存取。LiteLLM 會在將請求傳送至 Vertex AI 之前，自動將公開 URL 轉換為 base64 data URI。

區域可用性

Mistral OCR 可在多個區域使用。指定 vertex_location 以使用更接近您資料的區域：

• us-central1（預設）
• europe-west1
• asia-southeast1

Deepseek OCR 僅可在 global 區域使用。

支援的模型

• mistral-ocr-2505 - Vertex AI 上最新的 Mistral OCR 模型
• deepseek-ocr-maas - Vertex AI 上最新的 Deepseek OCR 模型

使用 Vertex AI 提供者前綴：vertex_ai/<model-name>