%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
Recently, IBM has launched four new Granite4.0Nano models, which have parameter sizes ranging from 3.5 million to 1.5 billion, demonstrating the company's emphasis on efficiency and accessibility in the field of artificial intelligence. Unlike large models in the industry that generally rely on cloud computing, the Granite4.0Nano models can run on regular laptops, and even locally in a browser. This allows developers to build applications for consumer hardware or edge devices without relying on cloud services.
All Granite4.0Nano models are released under the Apache 2.0 license, suitable for researchers, enterprises, and independent developers, and support commercial use. These models are compatible with various tools such as llama.cpp, vLLM, and MLX, and have been certified for responsible AI development under ISO42001.
The Granite4.0Nano family includes four models: Granite-4.0-H-1B (approximately 150 million parameters), Granite-4.0-H-350M (approximately 3.5 million parameters), Granite-4.0-1B (a variant with nearly 200 million parameters), and Granite-4.0-350M (a variant). The H-series models use a hybrid state space architecture, suitable for low-latency edge environments, while standard transformer models offer broader compatibility.
Although there are many small language models in the market, IBM's Nano family stands out due to its exceptional performance. Recent benchmark tests show that the Granite4.0 models perform better than competitors in multiple areas. IBM's research team stated that these models excel in instruction following and function calling tasks, and have advantages in memory requirements and runtime speed, allowing them to run smoothly on mobile devices or CPUs.
Notably, IBM not only released these models but also actively engaged with the developer community, collected feedback, and discussed future development plans. The Granite team communicated with users on Reddit, showcasing the company's efforts in promoting open-source AI development.
huggingface:https://huggingface.co/blog/ibm-granite/granite-4-nano
Key Points:
🌟 IBM released the Granite4.0Nano models, which support running on common hardware, enhancing the usability of small AI models.
🛠️ All models are released under the Apache 2.0 license, suitable for research and commercial use, ensuring openness and compatibility.
📈 Recent benchmark tests show that the Granite4.0 models outperform their competitors in performance, demonstrating efficient processing capabilities.