EschaLabs releases 2-bit quantized Qwen3.8-27B model for consumer GPUs

EschaLabs has released a 2-bit quantized version of the Qwen3.8-27B model, named Escha-W2. This new version significantly reduces the model's size to 10.15 GB, allowing it to fit on a single 24 GB consumer GPU with a 64k context window. Performance benchmarks indicate that Escha-W2 is competitive with its FP8 counterpart, even surpassing it in commonsense reasoning and matching it on LiveCodeBench, with a slight decrease in GPQA-Diamond performance. AI

IMPACT Enables running large language models on consumer hardware, potentially democratizing access and use.

RANK_REASON Release of a quantized model with performance benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Hugging Face Trending Models →

• EschaLabs/escha-runtime-qwen3dense
• EschaLabs/Qwen3.8-27B-Escha-W2
• Escha-W2
• huggingface_hub
• PyTorch
• Qwen3.8-27B
• Qwen3.8-27B-Escha-W2
• SGLang

• model release
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →