Preferred Networks launches PLaMo 3.1 Prime with Deep Research

Preferred Networks began offering PLaMo 3.1 Prime on October 27, 2026. The model is the latest Flagship in the company's Japanese-generated-AI PLaMo series and is available through PLaMo Chat, PLaMo API and on-premises deployment.

PLaMo 3.1 Prime adds web search and Deep Research features to PLaMo Chat and PLaMo API, allowing answers with sources based on the latest web information. Deep Research autonomously plans research, collects, organizes and analyzes information, and produces a sourced report over several minutes to tens of minutes.

The feature works in Cloud and on-premises environments. It can be combined with internal document search systems, or run only on internal documents in a closed environment.

Benchmark gains and a free API tier

Hallucination-related benchmark scores improved by an average of about 15% compared with PLaMo 3.0 Prime across eight benchmarks. A single model can now switch between reasoning and non-reasoning modes, and non-reasoning mode cuts output tokens by 90%. Inference Speed improved up to 1.7x through measures including speculative decoding.

A free plan for PLaMo API with usage caps and rate limits has also started. The paid Standard plan costs 60 yen per million input tokens and 250 yen per million output tokens.

PLaMo Chat has a redesigned UI and a limited-time campaign through November 30, 2026 that lets users try Deep Research free up to twice a day, according to forest.watch.impress.co.jp.