Gemini API without code: text, photos and strict JSON

You can use the Gemini API without code in nana800 with one block: paste a key from Google AI Studio, choose a Gemini model and write the task with data from earlier blocks. Gemini answers in text or in strict JSON that follows your schema, reads an image by link and compares two images. Each block run costs $0.002 in nana800, and Google counts the requests on your key by its own prices and limits.

Build it in the builder
  • Block run: $0.002
  • Free every month: $3.50
  • No card needed

What you can do

  • Reply to customers in a Telegram bot or by email with text written by Gemini
  • Get strict JSON that follows your schema and put the fields into a table or CRM
  • Read photos of receipts, documents and screenshots: describe them, find text, answer a question about the picture
  • Compare two images, for example a supplier photo and the product photo on your website
  • Summarize emails and leads, sort requests by topic
  • Pick a model for the task, from the cheap gemini-3.5-flash-lite to the smart gemini-4-argon

Blocks in the builder

  • Gemini request

    Action · $0.002 per run

    Sends text or images to the chosen Gemini model with your key. Three modes: answer a text prompt, analyze one image and compare two. The answer comes back as text or as JSON parsed into the json_data field.

How to set it up

  1. 1

    Get a Gemini API key

    Open Google AI Studio, sign in with your Google account and click Create API key. The key starts with AIzaSy. You can paste it straight into the block or keep it in the project variables.

  2. 2

    Create a scheme and add a trigger

    Create a scheme in the builder and choose what starts it: Telegram message from user to bot for a bot, webhook for website forms, or button to run it by hand and test.

  3. 3

    Add the Gemini AI request block

    Choose the mode, paste the key into the Gemini API key field and pick a model. In the prompt field write the task and insert data from the trigger. Use the system instruction for standing rules, for example: you are a polite support assistant, answer briefly.

  4. 4

    Set up images and JSON

    To read a photo, give a direct link to it: PNG, JPEG, WEBP, HEIC, HEIF or GIF up to 14 MB. To split the answer into fields, choose the json format and, if you like, add a JSON Schema so the answer always has the same structure.

  5. 5

    Pass the answer on

    Add a block that delivers the result: Telegram message send, send email message or google sheet record. The answer text is in text_response, the parsed JSON is in json_data.

  6. 6

    Turn the scheme on and test it

    Turn the scheme on, press the button trigger and check the answer in the run history of the scheme. The button itself is free. If it looks right, keep the scheme on. While the scheme is on, it costs $0.0042 per hour.

Frequently asked questions

A block run costs $0.002 whatever the answer length. Google counts the model requests on your key by its own prices and limits. An enabled scheme costs $0.0042 per hour. Your nana800 balance is topped up to $3.50 for free every month, no card needed. A scheme kept on around the clock for a month spends most of that on scheme hours.

In Google AI Studio: sign in with your Google account, click Create API key and copy the key that starts with AIzaSy. Paste it into the Gemini API key field of the block.

gemini-3.8-flash is the default: fast, inexpensive and good for most tasks. gemini-3.5-flash-lite is the cheapest. gemini-4-argon and gemini-3.1-pro-preview are the smartest, but slower and pricier. Old models that Google has shut down are replaced by the block with gemini-3.8-flash or gemini-3.5-flash-lite.

Choose json in the answer format field. Gemini returns strict JSON, and the parsed result appears in json_data, where the next blocks can read single fields. To keep the same set of fields on every run, add a JSON Schema.

PNG, JPEG, WEBP, HEIC, HEIF and GIF from a direct link that opens without a login. nana800 downloads the image itself and passes it to Gemini. The limit is 14 MB, and with two images it is 14 MB for both together.

The reason is in the error_message field. Common cases: a wrong key, a used up request limit, or Gemini declined the request under its safety rules. If the answer is empty, check the length limit: it also covers the reasoning of the model, and with a small number there is no room left for the answer.

Build your first scheme for free

Sign-up is free and needs no card. Every month the balance is topped up to $3.50 for free, enough for about 3,500 runs of basic blocks.

Build it in the builder