PLaMo releases Japanese-focused translation weights in four sizes
Preferred Networks released PLaMo 3 Translate in 0.8B, 2B, 8B and 31B sizes, offering a practical local translation option with conditional commercial licensing.
Preferred Networks published four downloadable PLaMo 3 Translate sizes on November 5, with BF16, quantized, GGUF and MLX variants.
The models support 53 languages, accept surrounding context and glossaries, and allow plain or polite Japanese. This makes specialist translation a deployment choice across very different memory budgets, rather than requiring every workflow to use a large general-purpose assistant. [1][2]
The company reports strong Japanese–English and Japanese-centered multilingual benchmark results, including for the smallest model. These are the developer’s evaluations, not a guarantee for a particular terminology, writing style or document type. A useful trial would compare translated passages against an approved glossary and human-reviewed reference examples from the intended business. [1][2]
Sources & details 2 sources
Discuss in your voice app
Take the reporting, visual explanations and original sources into a conversation. Copy the brief, paste it into your preferred assistant, then start its voice mode.
Paste the brief into your voice app to discuss this reporting.Reporting sources
- Preferred Networks — weight release
November 5 sizes, supported input controls, formats, vendor comparisons and commercial threshold.
- Preferred Networks — 0.8B model card
Downloadable specialist checkpoint and usage/licence guidance; not an independent translation evaluation.
Images & figures
PLaMo 3 Translate’s four sizes share a linear parameter scale. Model size affects deployment choices; these bars do not compare translation quality.
Euangelion · Source · Usage terms · Download figure