FINAL-Bench/Darwin-36B-KR

TEXT GENERATIONConcurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 23, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Darwin-36B-KR is a specialized Korean language model from the Darwin family, developed by FINAL-Bench. This model is optimized for Korean language tasks and is part of the POCKET series, designed for efficient on-device deployment. It is capable of running on resource-constrained environments like iPhones and CPU-only PCs using stock llama.cpp, making it highly accessible for localized applications.

Loading preview...

Darwin-36B-KR Overview

Darwin-36B-KR is a Korean-specialized language model developed by FINAL-Bench, part of their Darwin family of models. It is specifically designed to excel in tasks requiring deep understanding and generation of the Korean language. This model is a component of the POCKET series, which emphasizes efficient, on-device performance.

Key Capabilities

  • Korean Language Specialization: Optimized for high performance in Korean natural language processing tasks.
  • On-Device Deployment: Engineered to run efficiently on resource-limited hardware, including iPhones and CPU-only personal computers, utilizing stock llama.cpp without requiring specialized forks.
  • Accessibility: Provides a solution for localized applications where powerful Korean language capabilities are needed without reliance on cloud-based GPU infrastructure.

Good For

  • Korean Language Applications: Ideal for developers building applications that require robust Korean text generation, understanding, or translation.
  • Edge AI Scenarios: Suitable for use cases where models need to operate directly on user devices, ensuring privacy and reducing latency.
  • Resource-Constrained Environments: Excellent choice for projects targeting mobile devices or PCs without dedicated GPUs, offering powerful language capabilities in a compact footprint.

Finetunes
1 models