Doubt about Gemini API 2.0 vs 1.5

Gemini API
,
1

Hi making some tests in js, node.js i discovered the way to config a model and use it its some how:

 // Agrega el nuevo mensaje al historial
    historyGoogleai.push({
        role: "user",
        parts: [
            { text: userMessage }
        ]
    });

    const model = await genAI.getGenerativeModel({ model: chatbotSettings.chatbotModel });
    const result = await model.generateContent({
        contents: historyGoogleai,
        generationConfig: {
            temperature: 0.7,
            maxOutputTokens: 500,
        },
    });

My doubts;
1 - Gemini does not allow a system message as initial one, to describe the character, language, how to address to users…so the first message should be made by model…

2 - I like more to put the setup or config of model in getGenerativeAiModel, so you prepare one model with name of model-base, temperature, max_tokens etc… and once ready you only have to sen messages to that model…not every message to add the setup of the model… So its a way to use some kind of startChat and sendMessages in Gemini 2.0

3 - The same works for gemini in streaming no ?

4 - Offtopic: as Gemma is Ghttps://discuss.ai.google.dev/c/gemini-api/4emini open source, is there any link to download the different models
tHanks

3

Hello Eduardo,

I’m not a node developer so I won’t be able to answer all your questions but if you preferred the way the SDK or the APIs were working with 1.5, you can normally continue to do so as long as you don’t use 2.0 exclusive features like streaming.