Prism ML has released two versions of its Bonsai 27B model: one utilizing 1-bit binary weights and another with ternary weights. The 1-bit Bonsai model, optimized for mobile devices like the iPhone 17 Pro Max, boasts a significantly reduced footprint of approximately 3.9 GB while retaining about 90% of FP16 intelligence. The ternary version, designed for everyday laptops, offers a slightly larger footprint of around 7.2 GB but achieves 95% of FP16 intelligence and faster inference speeds. AI
IMPACT These highly compressed models could enable powerful AI capabilities on edge devices and standard laptops, expanding accessibility.
RANK_REASON The cluster details the release of new AI models by Prism ML, including specific versions and their technical specifications.
Read on Hugging Face Trending Models →
- Google Colab
- Jan
- Kaggle
- llama.cpp
- llama-cpp-python
- LM Studio
- Ollama
- prism-ml/Ternary-Bonsai-27B-gguf
- Ternary-Bonsai-27B-F16.gguf
- Unsloth Studio
- vLLM
- prism-ml/Bonsai-27B-gguf
- Hermes Agent
- Hugging Face
- Mlx
- OpenClaw
- prism-ml/Bonsai-27B-mlx-1bit
- prism-ml/Ternary-Bonsai-27B-mlx-2bit
AI-generated summary · Google Gemini · from 4 sources. How we write summaries →
