Last updated:
PLaMo
Preferred Networks' PLaMo — a Japanese language model developed from scratch in Japan, available by API or on-premises.
Model Facts
Latest model: PLaMo 3.0 Prime
- Provider
- Preferred Networks
- Released
- Jun 22, 2026
- Context window
- 256K tokens
- Weights
- Closed
Full Model List
| Model | Released | Status | Context | Note |
|---|---|---|---|---|
| PLaMo 3.0 Prime | Jun 22, 2026 | current | 256K tokens | Reasoning and Non-reasoning models, 256K context, OpenAI-compatible API, tool calling |
- Official model page:
- plamo.preferred.jp
- Other sources:
- PLaMo 3.0 Prime (Ledge.ai)
- PLaMo 3.0 Prime first look (Classmethod)
What it is
PLaMo 3.0 Prime, released on June 22, 2026, is Preferred Networks' flagship: a Japanese model developed from scratch and the first domestically built one with a Reasoning mode. It comes as Reasoning and Non-reasoning models with a 256K context window, an OpenAI-compatible API and tool calling, and can be used through the API or deployed on-premises. It builds on pre-training done jointly with NICT.
Where this information comes from
Details come from Japanese-language launch coverage. Release dates, context windows and benchmark scores are vendor-reported unless stated, and harnesses differ between labs, so treat the numbers as indicative.
Notable traits
- Strong Japanese-language performance.
- On-premises deployment option.
- Developed in Japan rather than fine-tuned from a foreign base model.