Use Gemma 4 for free with Google AI Studio

AI Studio runs Gemma 4 on Google's servers, so you need no download and no GPU.

1. 1Go to this AI Studio link and sign in. It opens a new chat with Gemma 4 31B.
2. 2Check that the model shown for the chat is Gemma 4 31B.
3. 3For faster replies, switch to the Gemma 4 26B A4B model in the model list.

You should see an empty chat, ready to use Gemma 4 31B.

31B is the larger dense model. 26B A4B only uses about 4 billion parameters per token, so it answers faster. Both read up to 256K tokens.

System instructions set the role, tone and format once, so you don't repeat them in every message.

1. 1Click Run settings in the top-right corner to open the panel.
2. 2Paste the text below into System Instructions, and fill in the part in brackets.
3. 3Leave temperature at 1.0, the value Google recommends for Gemma 4 in every task.

You are a helpful analyst for [YOUR ROLE OR TEAM]. Answer in plain English for a busy reader. Start with a one-line answer, then give up to 5 bullet points. When you read an image, say what you see before you explain it. If something is unclear or missing, say so instead of guessing.

You should see your instructions saved in the System Instructions field.

Gemma 4 reads images as well as text, so it can explain a chart or pull facts from a screenshot.

1. 1Use the add button next to the prompt box to upload your image.
2. 2Paste the prompt below into the Type something... box, then click Run.
3. 3Ask a follow-up in the same chat to dig into one detail.

What does this image show? List the 3 most important numbers or facts in it, and say what they mean for [YOUR GOAL].

You should see an answer in your format that names details from your image.

Google's tip for Gemma 4: put the image before the text. In AI Studio, add the image first, then type.

Get code turns the chat you built into an API request you can run anywhere.

1. 1When the chat gives the answers you want, click Get code.
2. 2Pick the language you work in and copy the code.

You should see code that calls the model gemma-4-31b-it.

An API key lets your own scripts and apps call Gemma 4, still for free.

1. 1Open the API keys page. New users already have a key there. Copy it, or click Create API key.
2. 2In a terminal, paste the command below with your key in place of [YOUR API KEY].
3. 3Press Enter to run it.

export GEMINI_API_KEY="[YOUR API KEY]"

curl "https://generativelanguage.googleapis.com/v1beta/models/gemma-4-31b-it:generateContent" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "system_instruction": {"parts": [{"text": "Answer in 2 short sentences."}]},
    "contents": [{"parts": [{"text": "What is an open model?"}]}]
  }'

You should see a JSON reply with Gemma's answer in the text field.

Keep your key out of code you share or commit. Store it in an environment variable, as here, or in a git-ignored file.