Lumma-0.6B-Base
A 600M-parameter multilingual text-generation model supporting English, Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Punjabi, and Odia. Features Shared KV caching for ~50% KV-cache memory reduction.
This is a base model (not instruction-tuned) — it performs text completion, not conversational chat. Give it a prompt and it will continue the text. Try prompts in English or any of the supported Indic languages.
Examples