{ "@context": "https://schema.org", "@id": "https://aggr.aymericbeaumet.com/items/philschmid-de/2026-04-07-how-to-use-gemma-4-with-the-gemini-api-and-google-ai-studio/#webpage", "@type": [ "WebPage", "ArchiveComponent" ], "articleBody": "Google's Gemma 4 family of open models is now available through the [Gemini API](https://ai.google.dev/gemma/docs/core/gemma_on_gemini_api) and [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemma-4-31b-it). Built from the same research behind Gemini 3, these models bring advanced reasoning, native function calling, multimodal understanding, and 256K context windows to an open, Apache 2.0-licensed package you can run anywhere.\n\nTwo models are available through the Gemini API today:\n\n- `gemma-4-26b-a4b-it`\n- `gemma-4-31b-it`\n\n## What makes Gemma 4 different\n\nGemma 4 models handle function calling, structured JSON output, and system instructions at the model level rather than through prompt engineering. The 31B dense model currently ranks as the #3 open model on the Arena AI text leaderboard, with the 26B MoE model at #6, competing with models 20x their size.\n\nKey capabilities:\n\n- 256K context window on both models\n- Native function calling and structured output\n- Multimodal text, images, and video\n- 140+ languages trained natively\n- Apache 2.0 license full commercial use, no restrictions\n\n## Getting started with AI Studio\n\nThe fastest way to try Gemma 4 is [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemma-4-31b-it). Select `gemma-4-26b-a4b-it` or `gemma-4-31b-it` from the model picker, type a prompt, and start chatting. You can test system instructions, adjust temperature, and experiment with multimodal inputs all through the browser. No API key or code required.\n\nOr click **Get Code** to export Python, JavaScript, or cURL snippets from any conversation.\n\n## Using Gemma 4 with the Gemini API\n\nInstall the Python SDK:\n\nBash\n\n```bash\npip install google-genai\n```\n\nSet your API key as an environment variable. You can create one at [aistudio.google.com/apikey](https://aistudio.google.com/apikey).\n\nBash\n\n```bash\nexport GEMINI_API_KEY=\"your-api-key\"\n```\n\n### Text generation\n\nGenerate text with Gemma 4:\n\nPython\n\n```python\nfrom google import genai\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"Compare ramen and udon in 3 bullet points: broth, noodle texture, and best season to eat.\"\n)\nprint(response.text)\n```\n\nPass a system instruction to set the model's behavior:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-31b-it\",\n config=types.GenerateContentConfig(\n system_instruction=\"You are a wise Kyoto tea master. Speak calmly and poetically, using nature metaphors. Keep answers under 3 sentences.\"\n ),\n contents=\"What is the purpose of the tea ceremony?\"\n)\nprint(response.text)\n```\n\n### Multi-turn conversations\n\nThe SDK provides a chat interface that tracks conversation history automatically:\n\nPython\n\n```python\nfrom google import genai\n \nclient = genai.Client()\nchat = client.chats.create(model=\"gemma-4-26b-a4b-it\")\n \nresponse = chat.send_message(\"What are the three most famous castles in Japan?\")\nprint(response.text)\n \nresponse = chat.send_message(\"Which one should I visit in spring for cherry blossoms?\")\nprint(response.text)\n```\n\n### Image understanding\n\nPass an image alongside your text prompt:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nwith open(\"path/to/image.png\", \"rb\") as f:\n image_bytes = f.read()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=[\n types.Part.from_bytes(data=image_bytes, mime_type=\"image/png\"),\n \"Describe this image in 2-3 sentences as if writing a caption for a Japanese travel magazine.\"\n ]\n)\nprint(response.text)\n```\n\n### Function calling\n\nDefine tools as function declarations. The model decides when to call them:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \n# Define the function declaration\nget_weather = {\n \"name\": \"get_weather\",\n \"description\": \"Get current weather for a given location.\",\n \"parameters\": {\n \"type\": \"object\",\n \"properties\": {\n \"location\": {\n \"type\": \"string\",\n \"description\": \"City and state, e.g. 'San Francisco, CA'\",\n },\n },\n \"required\": [\"location\"],\n },\n}\n \nclient = genai.Client()\ntools = types.Tool(function_declarations=[get_weather])\nconfig = types.GenerateContentConfig(tools=[tools])\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"Should I bring an umbrella to Kyoto today?\",\n config=config,\n)\n \n# The model returns a function call instead of text\nif response.candidates[0].content.parts[0].function_call:\n fc = response.candidates[0].content.parts[0].function_call\n print(f\"Function: {fc.name}\")\n print(f\"Arguments: {fc.args}\")\n```\n\n### Google Search\n\nGround Gemma 4 responses in real-time web data with Google Search:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"What are the dates for cherry blossom season in Tokyo this year?\",\n config=types.GenerateContentConfig(\n tools=[{\"google_search\":{}}]\n ),\n)\n \nprint(response.text)\n \n# Access grounding metadata for citations\nfor chunk in response.candidates[0].grounding_metadata.grounding_chunks:\n print(f\"Source: {chunk.web.title} — {chunk.web.uri}\")\n```\n\n## Where to go next\n\n- [Google AI Studio](https://aistudio.google.com) — Try Gemma 4 in the browser\n- [Gemini API documentation](https://ai.google.dev/gemini-api/docs)\n- [Gemma Documentation](https://ai.google.dev/gemma/docs)\n- [Gemma Cookbooks](https://github.com/google-gemma/cookbook)\n\n* * *\n\nThanks for reading! If you have any questions or feedback, please let me know on [Twitter](https://twitter.com/_philschmid) or [LinkedIn](https://www.linkedin.com/in/philipp-schmid-a6a2bb196/).\n", "authors": [], "captured_at": "2026-09-27T19:29:15.293681927+00:00", "category": "programming", "content_html": "

Google's Gemma 4 family of open models is now available through the Gemini API and Google AI Studio. Built from the same research behind Gemini 3, these models bring advanced reasoning, native function calling, multimodal understanding, and 256K context windows to an open, Apache 2.0-licensed package you can run anywhere.

\n

Two models are available through the Gemini API today:

\n\n

What makes Gemma 4 different

\n

Gemma 4 models handle function calling, structured JSON output, and system instructions at the model level rather than through prompt engineering. The 31B dense model currently ranks as the #3 open model on the Arena AI text leaderboard, with the 26B MoE model at #6, competing with models 20x their size.

\n

Key capabilities:

\n\n

Getting started with AI Studio

\n

The fastest way to try Gemma 4 is Google AI Studio. Select gemma-4-26b-a4b-it or gemma-4-31b-it from the model picker, type a prompt, and start chatting. You can test system instructions, adjust temperature, and experiment with multimodal inputs all through the browser. No API key or code required.

\n

Or click Get Code to export Python, JavaScript, or cURL snippets from any conversation.

\n

Using Gemma 4 with the Gemini API

\n

Install the Python SDK:

\n

Bash

\n
pip install google-genai\n
\n

Set your API key as an environment variable. You can create one at aistudio.google.com/apikey.

\n

Bash

\n
export GEMINI_API_KEY="your-api-key"\n
\n

Text generation

\n

Generate text with Gemma 4:

\n

Python

\n
from google import genai\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n    model="gemma-4-26b-a4b-it",\n    contents="Compare ramen and udon in 3 bullet points: broth, noodle texture, and best season to eat."\n)\nprint(response.text)\n
\n

Pass a system instruction to set the model's behavior:

\n

Python

\n
from google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n    model="gemma-4-31b-it",\n    config=types.GenerateContentConfig(\n        system_instruction="You are a wise Kyoto tea master. Speak calmly and poetically, using nature metaphors. Keep answers under 3 sentences."\n    ),\n    contents="What is the purpose of the tea ceremony?"\n)\nprint(response.text)\n
\n

Multi-turn conversations

\n

The SDK provides a chat interface that tracks conversation history automatically:

\n

Python

\n
from google import genai\n \nclient = genai.Client()\nchat = client.chats.create(model="gemma-4-26b-a4b-it")\n \nresponse = chat.send_message("What are the three most famous castles in Japan?")\nprint(response.text)\n \nresponse = chat.send_message("Which one should I visit in spring for cherry blossoms?")\nprint(response.text)\n
\n

Image understanding

\n

Pass an image alongside your text prompt:

\n

Python

\n
from google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nwith open("path/to/image.png", "rb") as f:\n    image_bytes = f.read()\n \nresponse = client.models.generate_content(\n    model="gemma-4-26b-a4b-it",\n    contents=[\n        types.Part.from_bytes(data=image_bytes, mime_type="image/png"),\n        "Describe this image in 2-3 sentences as if writing a caption for a Japanese travel magazine."\n    ]\n)\nprint(response.text)\n
\n

Function calling

\n

Define tools as function declarations. The model decides when to call them:

\n

Python

\n
from google import genai\nfrom google.genai import types\n \n# Define the function declaration\nget_weather = {\n    "name": "get_weather",\n    "description": "Get current weather for a given location.",\n    "parameters": {\n        "type": "object",\n        "properties": {\n            "location": {\n                "type": "string",\n                "description": "City and state, e.g. 'San Francisco, CA'",\n            },\n        },\n        "required": ["location"],\n    },\n}\n \nclient = genai.Client()\ntools = types.Tool(function_declarations=[get_weather])\nconfig = types.GenerateContentConfig(tools=[tools])\n \nresponse = client.models.generate_content(\n    model="gemma-4-26b-a4b-it",\n    contents="Should I bring an umbrella to Kyoto today?",\n    config=config,\n)\n \n# The model returns a function call instead of text\nif response.candidates[0].content.parts[0].function_call:\n    fc = response.candidates[0].content.parts[0].function_call\n    print(f"Function: {fc.name}")\n    print(f"Arguments: {fc.args}")\n
\n

Google Search

\n

Ground Gemma 4 responses in real-time web data with Google Search:

\n

Python

\n
from google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n    model="gemma-4-26b-a4b-it",\n    contents="What are the dates for cherry blossom season in Tokyo this year?",\n    config=types.GenerateContentConfig(\n        tools=[{"google_search":{}}]\n    ),\n)\n \nprint(response.text)\n \n# Access grounding metadata for citations\nfor chunk in response.candidates[0].grounding_metadata.grounding_chunks:\n    print(f"Source: {chunk.web.title} — {chunk.web.uri}")\n
\n

Where to go next

\n\n
\n

Thanks for reading! If you have any questions or feedback, please let me know on Twitter or LinkedIn.

\n", "content_markdown": "Google's Gemma 4 family of open models is now available through the [Gemini API](https://ai.google.dev/gemma/docs/core/gemma_on_gemini_api) and [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemma-4-31b-it). Built from the same research behind Gemini 3, these models bring advanced reasoning, native function calling, multimodal understanding, and 256K context windows to an open, Apache 2.0-licensed package you can run anywhere.\n\nTwo models are available through the Gemini API today:\n\n- `gemma-4-26b-a4b-it`\n- `gemma-4-31b-it`\n\n## What makes Gemma 4 different\n\nGemma 4 models handle function calling, structured JSON output, and system instructions at the model level rather than through prompt engineering. The 31B dense model currently ranks as the #3 open model on the Arena AI text leaderboard, with the 26B MoE model at #6, competing with models 20x their size.\n\nKey capabilities:\n\n- 256K context window on both models\n- Native function calling and structured output\n- Multimodal text, images, and video\n- 140+ languages trained natively\n- Apache 2.0 license full commercial use, no restrictions\n\n## Getting started with AI Studio\n\nThe fastest way to try Gemma 4 is [Google AI Studio](https://aistudio.google.com/prompts/new_chat?model=gemma-4-31b-it). Select `gemma-4-26b-a4b-it` or `gemma-4-31b-it` from the model picker, type a prompt, and start chatting. You can test system instructions, adjust temperature, and experiment with multimodal inputs all through the browser. No API key or code required.\n\nOr click **Get Code** to export Python, JavaScript, or cURL snippets from any conversation.\n\n## Using Gemma 4 with the Gemini API\n\nInstall the Python SDK:\n\nBash\n\n```bash\npip install google-genai\n```\n\nSet your API key as an environment variable. You can create one at [aistudio.google.com/apikey](https://aistudio.google.com/apikey).\n\nBash\n\n```bash\nexport GEMINI_API_KEY=\"your-api-key\"\n```\n\n### Text generation\n\nGenerate text with Gemma 4:\n\nPython\n\n```python\nfrom google import genai\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"Compare ramen and udon in 3 bullet points: broth, noodle texture, and best season to eat.\"\n)\nprint(response.text)\n```\n\nPass a system instruction to set the model's behavior:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-31b-it\",\n config=types.GenerateContentConfig(\n system_instruction=\"You are a wise Kyoto tea master. Speak calmly and poetically, using nature metaphors. Keep answers under 3 sentences.\"\n ),\n contents=\"What is the purpose of the tea ceremony?\"\n)\nprint(response.text)\n```\n\n### Multi-turn conversations\n\nThe SDK provides a chat interface that tracks conversation history automatically:\n\nPython\n\n```python\nfrom google import genai\n \nclient = genai.Client()\nchat = client.chats.create(model=\"gemma-4-26b-a4b-it\")\n \nresponse = chat.send_message(\"What are the three most famous castles in Japan?\")\nprint(response.text)\n \nresponse = chat.send_message(\"Which one should I visit in spring for cherry blossoms?\")\nprint(response.text)\n```\n\n### Image understanding\n\nPass an image alongside your text prompt:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nwith open(\"path/to/image.png\", \"rb\") as f:\n image_bytes = f.read()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=[\n types.Part.from_bytes(data=image_bytes, mime_type=\"image/png\"),\n \"Describe this image in 2-3 sentences as if writing a caption for a Japanese travel magazine.\"\n ]\n)\nprint(response.text)\n```\n\n### Function calling\n\nDefine tools as function declarations. The model decides when to call them:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \n# Define the function declaration\nget_weather = {\n \"name\": \"get_weather\",\n \"description\": \"Get current weather for a given location.\",\n \"parameters\": {\n \"type\": \"object\",\n \"properties\": {\n \"location\": {\n \"type\": \"string\",\n \"description\": \"City and state, e.g. 'San Francisco, CA'\",\n },\n },\n \"required\": [\"location\"],\n },\n}\n \nclient = genai.Client()\ntools = types.Tool(function_declarations=[get_weather])\nconfig = types.GenerateContentConfig(tools=[tools])\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"Should I bring an umbrella to Kyoto today?\",\n config=config,\n)\n \n# The model returns a function call instead of text\nif response.candidates[0].content.parts[0].function_call:\n fc = response.candidates[0].content.parts[0].function_call\n print(f\"Function: {fc.name}\")\n print(f\"Arguments: {fc.args}\")\n```\n\n### Google Search\n\nGround Gemma 4 responses in real-time web data with Google Search:\n\nPython\n\n```python\nfrom google import genai\nfrom google.genai import types\n \nclient = genai.Client()\n \nresponse = client.models.generate_content(\n model=\"gemma-4-26b-a4b-it\",\n contents=\"What are the dates for cherry blossom season in Tokyo this year?\",\n config=types.GenerateContentConfig(\n tools=[{\"google_search\":{}}]\n ),\n)\n \nprint(response.text)\n \n# Access grounding metadata for citations\nfor chunk in response.candidates[0].grounding_metadata.grounding_chunks:\n print(f\"Source: {chunk.web.title} — {chunk.web.uri}\")\n```\n\n## Where to go next\n\n- [Google AI Studio](https://aistudio.google.com) — Try Gemma 4 in the browser\n- [Gemini API documentation](https://ai.google.dev/gemini-api/docs)\n- [Gemma Documentation](https://ai.google.dev/gemma/docs)\n- [Gemma Cookbooks](https://github.com/google-gemma/cookbook)\n\n* * *\n\nThanks for reading! If you have any questions or feedback, please let me know on [Twitter](https://twitter.com/_philschmid) or [LinkedIn](https://www.linkedin.com/in/philipp-schmid-a6a2bb196/).\n", "dateCreated": "2026-09-27T19:29:15.293681927+00:00", "date_created": "2026-04-07T00:00:00+00:00", "date_modified": "2026-04-07T00:00:00+00:00", "date_published": "2026-04-07T00:00:00+00:00", "description": "Learn how to use Gemma 4 with the Gemini API and Google AI Studio.", "external_url": "https://www.philschmid.de/gemma-4-gemini-api", "git": { "commit": "ebb0941c5c06a2ab2377342d607535262a3e969e", "path": "items/philschmid-de/2026/04/2026-04-07-how-to-use-gemma-4-with-the-gemini-api-and-google-ai-studio.md", "permalink": "https://github.com/aymericbeaumet/aggr-instance/blob/ebb0941c5c06a2ab2377342d607535262a3e969e/items/philschmid-de/2026/04/2026-04-07-how-to-use-gemma-4-with-the-gemini-api-and-google-ai-studio.md" }, "id": "urn:aggr:item:f020c2a21d1e9c228316728ff1b21e7d2f9a9fb3", "isBasedOn": { "@id": "https://www.philschmid.de/gemma-4-gemini-api" }, "mainEntity": { "@id": "https://www.philschmid.de/gemma-4-gemini-api", "@type": "CreativeWork", "archivedAt": { "@id": "https://aggr.aymericbeaumet.com/items/philschmid-de/2026-04-07-how-to-use-gemma-4-with-the-gemini-api-and-google-ai-studio/#webpage" }, "dateModified": "2026-04-07T00:00:00+00:00", "datePublished": "2026-04-07T00:00:00+00:00", "headline": "How to use Gemma 4 with the Gemini API and Google AI Studio", "timeRequired": "PT4M", "url": "https://www.philschmid.de/gemma-4-gemini-api", "wordCount": 706 }, "name": "Archived snapshot: How to use Gemma 4 with the Gemini API and Google AI Studio", "preview": null, "reading_minutes": 4, "replicated_at": null, "source": "philschmid-de", "source_name": "philschmid.de - RSS feed", "summary": "Learn how to use Gemma 4 with the Gemini API and Google AI Studio.", "tags": [], "temporalCoverage": "2026-09-27T19:29:15.293681927+00:00", "title": "How to use Gemma 4 with the Gemini API and Google AI Studio", "url": "https://aggr.aymericbeaumet.com/items/philschmid-de/2026-04-07-how-to-use-gemma-4-with-the-gemini-api-and-google-ai-studio/", "word_count": 706 }