Google launches Gemma 4 AI model for developers

Search engine giant Google has launched the new Gemma 4, touted to be the company's most advanced and powerful generative Artificial Intelligence (gen AI) Large Language Model (LLM).

Gemma AI model is popular among software developers to build their own custom AI applications. As it is an open model, they can fine-tune it and quantise it as per the requirements. Since the launch of the first-generation Gemma AI model, developers have downloaded it over 400 million times. So far, they have built more than 100,000 variants.

Google says Gemma 4 delivers an unprecedented level of intelligence-per-parameter. It is offering it in four models-- Effective 2B (E2B), Effective 4B (E4B), 26B Mixture of Experts (MoE) and 31B Dense. All the variants deliver beyond simple chat, can handle complex logic and agentic workflows.

Gemma 4 models are designed to specifically to run and fine-tune efficiently on hardware, such as Android devices, laptop GPUs, developer workstations and accelerators, as well.

Gemma models are capable of supporting multi-step planning and deep logic, and also come with significant improvements in solving complex mathematical problems.

Also, Gemma 4 supports high-quality offline code, which can turn a workstation into a local-first AI code assistant.

Gemma 4 also natively supports more than 140 languages globally.

Google's Gemma 4 can be downloaded and tested on Google AI Studio and even from third parties such as Hugging Face, Kaggle, or Ollama websites.

Get the latest news on new launches, gadget reviews, apps, cybersecurity, and more on personal technology only on DH Tech.