playing-around-with-gemini-ai-api

Wanted to move from OpenAi API to gemini-pro API. That's why I started to playing around with it to see how it goes with my need. So far I'm very pleased.

Downside

The only downside is JSON object generation by the model. gemini-pro generates parsable JSON string inside markdown syntax. Probably in training, it sees a lot JSON data in markdown documentation. So it respond with it. JSON generation was a bit problem in OpenAI too. But with some checks, validations or rerequesting, those can be overcome. Basically we are in the bleeding edge of AI era and AIs are still learning and improving ...fast.

Currenlty there are some unsupported regions from where Gemini API can not be called/used.

List of supported countries and languages: https://ai.google.dev/available_regions

If your app is hosted in some unsupported countries like (as of Mar 2024) most of the European countries, Hong Kong etc., your API requests will be failed.

Sample Response Object

Success

{
   "response": {
      "candidates": [
         {
            "content": {
               "parts": [
                  {
                     "text": "```jsonn{\\n  \\"title\\": \\"The Importance of Questioning\\",\\n  \\"types\\": [\\n    \\"Philosophy\\",\\n    \\"Critical Thinking\\"\\n  ],\\n  \\"categories\\": [\\n    \\"Reasoning\\",\\n    \\"Decision Making\\",\\n    \\"Problem Solving\\"\\n  ],\\n  \\"keywords\\": [\\n    \\"Questioning\\",\\n    \\"Answering\\",\\n    \\"Conclusion\\",\\n    \\"Reasoning\\",\\n    \\"Critical Thinking\\"\\n  ],\\n  \\"overallRating\\": 8,\\n  \\"clarityRating\\": 9,\\n  \\"criticism\\": \\"The article is well-written and easy to understand. It provides a clear and concise overview of the importance of questioning. However, the article could be improved by providing more specific examples of how questioning can be used to improve decision-making and problem-solving.\\"\\n}\\n```"
                  }
               ],
               "role": "model"
            },
            "finishReason": "STOP",
            "index": 0,
            "safetyRatings": [
               {
                  "category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
                  "probability": "NEGLIGIBLE"
               },
               {
                  "category": "HARM_CATEGORY_HATE_SPEECH",
                  "probability": "NEGLIGIBLE"
               },
               {
                  "category": "HARM_CATEGORY_HARASSMENT",
                  "probability": "NEGLIGIBLE"
               },
               {
                  "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
                  "probability": "NEGLIGIBLE"
               }
            ]
         }
      ],
      "promptFeedback": {
         "safetyRatings": [
            {
               "category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
               "probability": "NEGLIGIBLE"
            },
            {
               "category": "HARM_CATEGORY_HATE_SPEECH",
               "probability": "NEGLIGIBLE"
            },
            {
               "category": "HARM_CATEGORY_HARASSMENT",
               "probability": "NEGLIGIBLE"
            },
            {
               "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
               "probability": "NEGLIGIBLE"
            }
         ]
      }
   }
}

Sample REST request

curl https://generativelanguage.googleapis.com/v1beta/models/gemini-pro:generateContent?key=$API_KEY \
    -H 'Content-Type: application/json' \
    -X POST \
    -d '{
        "contents": [{
            "parts":[
                {"text": "Write a story about a magic backpack."}
            ]
        }],
        "safetySettings": [
            {
                "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
                "threshold": "BLOCK_ONLY_HIGH"
            }
        ],
        "generationConfig": {
            "stopSequences": [
                "Title"
            ],
            "temperature": 1.0,
            "maxOutputTokens": 800,
            "topP": 0.8,
            "topK": 10
        }
    }'  2> /dev/null | grep "text"