VIDRAFT has released POCKET-Darwin-180B-GGUF, a quantized version of their 180B parameter Mixture-of-Experts model designed to run on consumer hardware. This version, available in GGUF format, can operate on a gaming laptop with 8GB VRAM and 32GB RAM by utilizing sparse activation and SSD streaming via llama.cpp. The model reportedly maintains its accuracy after 4-bit quantization, with no measurable degradation on the MMLU-Pro benchmark according to VIDRAFT's self-reported tests. AI
IMPACT Enables deployment of large-scale models on consumer hardware, potentially broadening access for on-premise and air-gapped applications.
RANK_REASON Release of a quantized, consumer-deployable version of a large MoE model.
- Darwin-180B-RSI
- Hugging Face
- llama.cpp
- modelscope
- POCKET-Darwin-180B-GGUF
- Qwen3.8 Flash-Next
- VIDRAFT
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →