Gemma-3-12B
A lightweight, state-of-the-art multimodal model from Google featuring advanced reasoning, image understanding, and a massive 128K context window.
Total Parameters
12B
Context Window
128K
Tokens
Modalities
Text & Image
Max Output
8,192
Tokens
Accessible Multimodal Innovation
Gemma 3 12B is a state-of-the-art open multimodal model from Google, built upon the same pioneering research as the Gemini models. As the second largest in the Gemma 3 family, it is trained on 12 trillion tokens and handles both text and image inputs to generate high-quality text output. Featuring an expansive 128K context window, native support for over 140 languages, and advanced function calling, its optimized size allows it to be deployed efficiently on diverse infrastructures while democratizing access to cutting-edge AI.
Multimodal 12B — Text & Image
Built for accessible multimodal intelligence
Image Data Extraction
Extract, interpret, and summarize complex visual data from images (normalized to 896 × 896 resolution) for seamless text-based communication and analysis.
Conversational AI & Creation
Power conversational interfaces, virtual assistants, and chatbots, or generate creative formats like code, marketing copy, and structured outputs.
Long-Form Summarization
Leverage the massive 128K context window to ingest research papers, extensive reports, and large text corpora, generating concise and accurate summaries.
Research & Education
Serve as a foundational tool for NLP/VLM research, build interactive language learning experiences across 140+ languages, and assist in deep knowledge exploration.
Multimodal Capabilities
Baseline performance across reasoning, coding, and agentic workflows for the 12B weight class.
Intelligence Index
Better than 11% of models
GPQA Diamond
Better than 28% of models
τ²-Bench Telecom
Better than 18% of models
| Category | Benchmark | Score | Description |
|---|---|---|---|
| Reasoning | GPQA Diamond | 35% | Graduate-level scientific reasoning |
| Reasoning | τ²-Bench Telecom | 11% | AI agents in dual-control scenarios |
| Reasoning | IFBench | 37% | Instruction-following accuracy |
| Coding | SciCode | 17% | Python for scientific computing |
| Knowledge | AA-Omniscience Accuracy | 10% | Proportion of correctly answered questions |
Metrics sourced from Artificial Analysis. Hallucination Rate: 2.9%
Flexible Pricing Tiers
Choose the optimal balance of speed and cost for your workflow. Prices are per 1M tokens.
| Tier | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
| Standard (Overnight) | $0.02 | $0.20 |
| Async | $0.03 | $0.30 |
Context window natively supported up to 128k tokens.
Start Building in Minutes
Gemma-3-12B is accessible via OpenAI-compatible endpoints. Here is how to integrate it using the standard Python SDK via Doubleword.ai.
Developer Tip: Multimodal Inputs
Gemma 3 supports vision-language inputs. When passing images to the model, they are normalized to 896 × 896 resolution and encoded to 256 tokens each within your request payload.
from openai import OpenAI
client = OpenAI(
api_key="your-api-key-here",
base_url="https://api.doubleword.ai/v1"
)
# Step 1: Upload a batch input file
with open("batch_requests.jsonl", "rb") as file:
batch_file = client.files.create(
file=file,
purpose="batch"
)
print(f"File ID: {batch_file.id}")
# Step 2: Create a batch job
batch = client.batches.create(
input_file_id=batch_file.id,
endpoint="/v1/chat/completions",
completion_window="24h"
)
print(f"Batch ID: {batch.id}")
# Step 3: Check batch status
batch_status = client.batches.retrieve(batch.id)
print(f"Status: {batch_status.status}")