TranslateGemma 4B
TranslateGemma 4B is a translation-specialized foundation model developed by Google DeepMind. It has a parameter scale of approximately 5 billion and supports a long context window of 128K.
Parameters
5.0B
Context Window
128K
License
Gemma License
Release Date
2026-01-15
API Pricing
API pricing for this model is not yet available
Strengths
- ・High performance specialized in translation
- ・128K broad context capacity
- ・Efficient 5B parameter scale
Weaknesses
- ・General tasks beyond translation
- ・10GB model file size
- ・Subject to limited usage terms
Use Cases
- ・Multilingual translation of long documents
- ・Translation processing of specialized text
- ・Building efficient translation pipelines
Deep Analysis
Parameters
4B
Smallest TranslateGemma variant, suitable for mobile inference
Languages
55 core / ~500 extended
Core: high/mid/low-resource. Extended: nearly 500 additional pairs
Performance
Rivals Gemma 3 12B baseline
4B model matches the translation quality of the 12B base model
Base Model
Gemma 3
Distilled from Gemini models via SFT + RL
Release Date
January 15, 2026
Announced by Google
Multimodal
Yes
Retains Gemma 3 multimodal capabilities for image text translation
License
Gemma license
Available on Kaggle and Hugging Face
Training
SFT + RL
Supervised fine-tuning on parallel data + reinforcement learning with MetricX-QE and AutoMQM reward models
Strengths
- ・Rivals Gemma 3 12B baseline translation quality at 1/3 the parameters
- ・Supports 55 core languages and ~500 extended language pairs
- ・Multimodal: can translate text within images (Vistra benchmark)
- ・Runs on mobile hardware - ideal for on-device translation
- ・Distilled from Gemini models for high quality
- ・Open weights, available on Kaggle and Hugging Face
Weaknesses
- ・Lower quality than 12B and 27B TranslateGemma variants
- ・Limited to translation tasks (not a general-purpose model)
- ・Extended language pairs (~500) lack confirmed evaluation metrics
- ・May struggle with very complex or domain-specific translations
- ・Not as widely benchmarked as general-purpose models
Competitor Comparison
| Model | Arena | SWE | GPQA | Price |
|---|---|---|---|---|
| TranslateGemma 4B | N/A | N/A | N/A | Free (open weights) |
| TranslateGemma 12B | N/A | N/A | N/A | Free (open weights) |
| TranslateGemma 27B | N/A | N/A | N/A | Free (open weights) |
| NLLB-200 (3.3B) | N/A | N/A | N/A | Free (open weights) |
| Gemma 3 27B (baseline) | ~1430 | N/A | ~70% | Free (open weights) |
TranslateGemma 4B is the smallest variant in Google's open translation model family, rivaling the Gemma 3 12B baseline's translation quality with only 4B parameters. Supports 55 core languages with multimodal image text translation. Ideal for mobile and edge deployment.
Sources
Analysis generated: 2026-05-24