Google Open-Sources Gemma 3 270M, Beats Qwen 2.5 in Performance
Google Gemma 3 270M: Tiny 241MB AI model with 270M params, outperforms Qwen 2.5. Energy-efficient, instruction-tuned for on-device & specialized tasks.
"AI Disruption" Publication 7400 Subscriptions 20% Discount Offer Link.
The download is only 241 MB.
This Thursday, Google officially released the latest model in the Gemma 3 series.
Gemma 3 270M is a compact, small-scale language model with 270 million parameters, specifically designed for task-specific fine-tuning, with strong instruction-following and text structuring capabilities.
It inherits the advanced architecture and powerful pre-training capabilities of the Gemma 3 series, while bringing robust instruction execution abilities to small-scale models. As demonstrated by Google's IFEval benchmark results, Gemma 3 270M sets a new performance standard among models in its class, making complex AI functionality more accessible for on-device and research applications.
