(WIP) [Mistral] - Add dedicated OCR notebook (#4088)

* Add dedicated OCR notebook

* - Remove OCR mentions in initial notebook
- Clear outputs and variable names in OCR notebook

* Fix

* Fix

Author: hrjn

notebooks/official/generative_ai/mistralai_intro.ipynb

@@ -85,9 +85,6 @@
         "\n",
         "### Available Mistral AI models\n",
         "\n",
-        "*   ### Mistral OCR (25.05)\n",
-        "Mistral OCR (25.05) is a model specialized in extracting text and images from documents. It is specifically built to preserve the structure of the document pages and automatically formats the extracted text in Markdown.\n",
-        "\n",
         "*   ### Mistral Small 3.1 (25.03)\n",
         "Mistral Small 3.1 (25.03) is the enhanced version of Mistral Small 3, featuring multimodal capabilities and an extended context length of up to 128k.\n",
         "\n",
@@ -174,16 +171,13 @@
       },
       "outputs": [],
       "source": [
-        "MODEL = \"mistral-small-2503\"  # @param [\"mistral-small-2503\", \"codestral-2501\", \"mistral-large-2411\", \"mistral-nemo\", \"mistral-ocr-2505\"]\n",
+        "MODEL = \"mistral-small-2503\"  # @param [\"mistral-small-2503\", \"codestral-2501\", \"mistral-large-2411\", \"mistral-nemo\"]\n",
         "if MODEL == \"mistral-small-2503\":\n",
         "    available_regions = [\"europe-west4\", \"us-central1\"]\n",
         "    available_versions = [\"latest\"]\n",
         "elif MODEL == \"mistral-large-2411\":\n",
         "    available_regions = [\"europe-west4\", \"us-central1\"]\n",
         "    available_versions = [\"latest\"]\n",
-        "elif MODEL == \"mistral-ocr-2505\":\n",
-        "    available_regions = [\"europe-west4\", \"us-central1\"]\n",
-        "    available_versions = [\"latest\"]\n",
         "elif MODEL == \"mistral-nemo\":\n",
         "    available_regions = [\"europe-west4\", \"us-central1\"]\n",
         "    available_versions = [\"2407\"]\n",
@@ -300,15 +294,6 @@
         "import requests"
       ]
     },
-    {
-      "cell_type": "markdown",
-      "metadata": {
-        "id": "Y_8xeTfWqToW"
-      },
-      "source": [
-        "To try the OCR model, please refer to the section \"Performing OCR\"."
-      ]
-    },
     {
       "cell_type": "markdown",
       "metadata": {
@@ -1413,98 +1398,6 @@
         "    print(f\"Request failed with status code: {response.status_code}\")"
       ]
     },
-    {
-      "cell_type": "markdown",
-      "metadata": {
-        "id": "znQtFeXUqToe"
-      },
-      "source": [
-        "### Performing OCR\n",
-        "\n",
-        "Mistral OCR (`mistral-ocr-2505`) is an Optical Character Recognition API that sets a new standard in document understanding. Unlike other models, Mistral OCR comprehends each element of documents: images, text, tables, equations, etc. It takes images and PDFs as input and extracts content in an ordered interleaved text and images.\n",
-        "\n",
-        "The following example showcases the application of the OCR API on a PDF file."
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "id": "KRaYEGJ1qToe"
-      },
-      "outputs": [],
-      "source": [
-        "import base64\n",
-        "\n",
-        "# The URL you provided\n",
-        "pdf_url = \"https://arxiv.org/pdf/2501.12948\"\n",
-        "\n",
-        "# 1. Download the file\n",
-        "print(f\"Attempting to download from: {pdf_url}\")\n",
-        "response = requests.get(pdf_url)\n",
-        "pdf_content_bytes = response.content\n",
-        "\n",
-        "# 2. Base64 encode the downloaded content\n",
-        "encoded_content_bytes = base64.b64encode(pdf_content_bytes)\n",
-        "encoded_string = encoded_content_bytes.decode(\"utf-8\")\n",
-        "\n",
-        "print(\"Download and encoding complete.\")\n",
-        "\n",
-        "pdf_doc_base64_uri = f\"data:application/pdf;base64,{encoded_string}\"\n",
-        "\n",
-        "# print(pdf_doc_base64_uri)"
-      ]
-    },
-    {
-      "cell_type": "code",
-      "execution_count": null,
-      "metadata": {
-        "id": "Tn3sECpRqToj"
-      },
-      "outputs": [],
-      "source": [
-        "# Get the access token\n",
-        "process = subprocess.Popen(\n",
-        "    \"gcloud auth print-access-token\", stdout=subprocess.PIPE, shell=True\n",
-        ")\n",
-        "(access_token_bytes, err) = process.communicate()\n",
-        "access_token = access_token_bytes.decode(\"utf-8\").strip()  # Strip newline\n",
-        "\n",
-        "OCR_MODEL = \"mistral-ocr-2505\"\n",
-        "url = f\"{ENDPOINT}/v1/projects/{PROJECT_ID}/locations/{LOCATION}/publishers/mistralai/models/{OCR_MODEL}:rawPredict\"\n",
-        "data = {\n",
-        "    \"model\": OCR_MODEL,\n",
-        "    \"document\": {\"type\": \"document_url\", \"document_url\": pdf_doc_base64_uri},\n",
-        "}\n",
-        "headers = {\n",
-        "    \"Authorization\": f\"Bearer {access_token}\",\n",
-        "    \"Content-Type\": \"application/json\",\n",
-        "}\n",
-        "\n",
-        "# Make the POST request\n",
-        "response = requests.post(url, headers=headers, json=data)\n",
-        "\n",
-        "# Check status code and try to parse the response as JSON\n",
-        "if response.status_code == 200:\n",
-        "    try:\n",
-        "        response_dict = response.json()\n",
-        "        print(response_dict)\n",
-        "    except json.JSONDecodeError as e:\n",
-        "        print(\"Error decoding JSON:\", e)\n",
-        "        print(\"Raw response:\", response.text)  # Print raw response if parsing fails\n",
-        "else:\n",
-        "    print(f\"Request failed with status code: {response.status_code}\")"
-      ]
-    },
-    {
-      "cell_type": "markdown",
-      "metadata": {
-        "id": "CmpLW-IEqToj"
-      },
-      "source": [
-        "To get more details on the options available when calling the OCR API, please refer to the [Mistral API documentation](https://docs.mistral.ai/api/#tag/ocr)."
-      ]
-    },
     {
       "cell_type": "markdown",
       "metadata": {