hex-1-f32-GGUF

Hex1 is a 4B parameter, open-source language model optimized for Indian languages, designed to address India’s linguistic diversity by enabling intelligent systems that natively understand and respond in Hindi, Kannada, Telugu, Tamil, and Malayalam. Built on the Qwen/Qwen3-4B-Base, Hex1 outperforms leading models such as Gemma-2B, LLaMA-3.2-3B, and Sarvam-1 on key benchmarks like MMLU, making it one of the most capable options for Indic language tasks. The model supports commercial licensing to empower businesses and developers to build applications without restrictive usage terms, and promises future expansion to cover more Indian languages. With robust multilingual performance and accessible open-source availability, Hex1 aims to include the 90% of India’s population underserved by English-centric AI, unlocking the potential of Generative AI across India’s rich linguistic landscape.

Model Files

File Name	Quant Type	File Size
hex-1.BF16.gguf	BF16	8.05 GB
hex-1.F16.gguf	F16	8.05 GB
hex-1.F32.gguf	F32	16.1 GB
hex-1.Q2_K.gguf	Q2_K	1.67 GB
hex-1.Q3_K_L.gguf	Q3_K_L	2.24 GB
hex-1.Q3_K_M.gguf	Q3_K_M	2.08 GB
hex-1.Q3_K_S.gguf	Q3_K_S	1.89 GB
hex-1.Q4_K_M.gguf	Q4_K_M	2.5 GB
hex-1.Q4_K_S.gguf	Q4_K_S	2.38 GB
hex-1.Q5_K_M.gguf	Q5_K_M	2.89 GB
hex-1.Q5_K_S.gguf	Q5_K_S	2.82 GB
hex-1.Q6_K.gguf	Q6_K	3.31 GB
hex-1.Q8_0.gguf	Q8_0	4.28 GB

Quants Usage

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):