How to Access Gemma-3-12B-IT in 3 Ways?

Gemma-3-12B-IT belongs to Google’s Gemma family of open models and delivers strong multimodal performance while remaining lightweight and efficient. Built on the same advanced foundation as Gemini, it handles tasks such as text generation, summarization, reasoning, and image understanding with ease, offering a powerful yet accessible option for developers and researchers alike.

In this guide, we will begin with a quick review of Gemma-3-12B-IT and then walk through different ways to access it, including web interfaces, API integration, and local deployment.

What is Gemma-3-12B-IT?

Basic Information

Feature	Details
Model Size	12B Parameters
Architecture	Dense
Open Source	Yes
Context Window	128K Tokens
Multilingual Support	Excels in English, support 140 languages
Multimodality	Text and Images (normalized to 896 x 896 resolution)
License	Gemma

Benchmark	Performance
GPQA Diamond	35%
MMLU-Pro	60%
IFBench	37%
SciCode	17%
LiveCodeBench	14%
AIME 2025	18%
Humanity’s Last Exam	4.8%
AA-LCR	7%

Extended Context Processing

With a 128,000-token context window, Gemma-3-12B-IT moves beyond a mere technical upgrade as it redefines how organizations process lengthy documents and intricate analytical workflows. Its advanced design removes the fragmentation issues found in conventional models, allowing seamless comprehension across large volumes of text without losing coherence or context.

This expanded capacity opens new frontiers for document intelligence, letting AI systems retain understanding throughout entire research papers, contracts, or technical manuals while also interpreting visual components such as graphs, charts, and illustrations.

Advanced Multimodal Integration

Built with a vision-language framework, Gemma-3-12B-IT advances far beyond standard image recognition to achieve human-like analytical reasoning. By linking textual and visual information, it can interpret relationships between the two modalities and extract deeper insights that would be inaccessible through text-only or image-only analysis.

Key Highlights

• Document Analysis: Pull out useful insights from reports that include charts, graphs, and visuals.
• Visual Understanding: Answer complex image-based questions with clear and logical reasoning.
• Content Generation: Write clear descriptions, captions, and explanations that connect visuals and text naturally.
• Learning Support: Offer thorough, easy-to-grasp explanations that combine text with helpful visual examples.

Instruction-Tuned Architecture

Gemma-3-12B-IT’s refined instruction-tuning design streamlines the AI deployment process by minimizing the need for complex prompt engineering or advanced technical setup. It naturally interprets human language commands and preserves context through extended, multi-turn conversations, enabling smoother and more intuitive interaction with the model.

How to Access Gemma-3-12B-IT: Web Interface (for Beginner)

How to Access Gemma-3-12B-IT: Using the API (for Developer)

Novita AI provides Gemma-3-12B-IT API with 131K context, and costs of $0.05/input and $0.1/output, allowing developers to seamlessly tap into Google’s lightweight multimodal model for advanced reasoning, summarization, and generation tasks via one unified API.

Novita AI

Step 1: Log In and Access the Model Library

Log in to your account and click on the Model Library button.

Step 2: Start Your Free Trial

Select your modal and begin your free trial to explore the capabilities of the selected model.

Step 3: Get Your API Key

To authenticate with the API, we will provide you with a new API key. Entering the “Settings“ page, you can copy the API key as indicated in the image.

Step 4: Install the API

Install API using the package manager specific to your programming language.

After installation, import the necessary libraries into your development environment. Initialize the API with your API key to start interacting with Novita AI LLM. This is an example of using chat completions API for python users.

from openai import OpenAI
  
client = OpenAI(
    base_url="https://api.novita.ai/openai",
    api_key="session_Um3Ozta39g2J__yeP9b_rOegzeA_qSYYquKzJS2oitKENIo8_H2FL2sCtl25-sKWjCY_wsmN18iuDp1zv_Xkaw==",
)

model = "google/gemma-3-12b-it"
stream = True # or False
max_tokens = 4096
system_content = "Be a helpful assistant"
temperature = 1
top_p = 1
min_p = 0
top_k = 50
presence_penalty = 0
frequency_penalty = 0
repetition_penalty = 1
response_format = { "type": "text" }

chat_completion_res = client.chat.completions.create(
    model=model,
    messages=[
        {
            "role": "system",
            "content": system_content,
        },
        {
            "role": "user",
            "content": "Hi there!",
        }
    ],
    stream=stream,
    max_tokens=max_tokens,
    temperature=temperature,
    top_p=top_p,
    presence_penalty=presence_penalty,
    frequency_penalty=frequency_penalty,
    response_format=response_format,
    extra_body={
      "top_k": top_k,
      "repetition_penalty": repetition_penalty,
      "min_p": min_p
    }
  )

if stream:
    for chunk in chat_completion_res:
        print(chunk.choices[0].delta.content or "", end="")
else:
    print(chat_completion_res.choices[0].message.content)

Option 2: Multi-Agent Workflows with OpenAI Agents SDK

Build sophisticated multi-agent systems leveraging Gemma-3-12B-IT’s dual-mode capabilities:

• Plug-and-Play Integration: Use DeepSeek V3.1 in any OpenAI Agents workflow
• Advanced Agent Capabilities: Support for handoffs, routing, and tool integration
• Scalable Architecture: Design agents that leverage DeepSeek V3.1’s capabilities

How to Access Gemma-3-12B-IT: Local Deployment (for Advanced Users)

Gemma3-12B-IT Hardware Requirements

Quantization	Weights Only (Approx.)	With KV-cache (Approx.)	Minimum Configuration	Recommended GPU
BF16	24.0 GB	38.9 GB	Nvidia L40S ×1	Nvidia H100 ×1
SFP8	12.4 GB	27.3 GB	Nvidia L40S ×1	Nvidia A100 ×1
INT4	6.6 GB	21.5 GB	Nvidia L4 ×1	Nvidia L40S ×1

For users seeking greater control and flexibility, Novita AI provides on-demand Cloud GPU instances includeing L40S, A100, H100, as well as other demanding options such as RTX 4090, RTX 5090, and RTX 6000 Ada, allowing users to deploy high-performance workloads effortlessly without relying on local hardware.

Best Practices for Using Gemma-3-12B-IT

• Choose the Right Access Method: Beginners can start with the web interface for quick trials, while developers should use the Novita AI API for integration into apps and workflows. Advanced users may prefer local deployment for full control and offline use.
• Mind Resource Requirements: If deploying locally, confirm your GPU meets the minimum configuration—quantized models such as INT4 or SFP8 are ideal for balancing performance and memory efficiency.
• Optimize for Context and Throughput: Gemma-3-12B-IT supports up to 128K tokens. For longer inputs, split content into structured segments or use summarization to maintain coherent results.
• Leverage Multimodal Strengths: Combine text and images in prompts to explore the model’s analytical reasoning and descriptive generation capabilities.
• Experiment and Iterate: Adjust parameters like temperature, top_p, and max_tokens to fine-tune creativity, factuality, and response length according to your task.

Frequently Asked Questions

Novita AI is a leading AI cloud platform that provides developers with easy-to-use APIs and affordable, reliable GPU infrastructure for building and scaling AI applications.