Gemma 4
Gemma 4 is Google DeepMind's current family of lightweight, open-weight language models, announced in April 2026 as the successor to earlier Gemma generations. It's designed to run across a range of hardware, from cloud servers down to laptops and phones, and is distributed through Hugging Face, Kaggle and Google's own tooling. Like earlier Gemma models, the weights are free to download and use under Google's own Gemma license terms.
Facts checked on against the sources listed at the end of this page.
- Pricing
- Open sourceFree plan: Yes, weights are free to download and run
- Open source
- NoGemma license (open weights, with Google's own usage terms)
- GitHub stars
- 5,766
- Latest release
- v4.0.14 months ago
A good fit for
- Developers who want a free, open-weight model from Google to self-host
- Running a capable language model on modest hardware, including laptops
- Teams that want to fine-tune an open model for a specific task
- Projects that want the option to move between self-hosting and Google's managed Vertex AI
Not a good fit for
- Applications that need Google's absolute top proprietary model, since Gemma is the lightweight, open-weight line rather than Google's largest closed models
- Teams that need a fully permissive open source license with no usage restrictions, since Gemma ships under Google's own license terms
- Very large-scale production inference where a managed API might be simpler than self-hosting, unless already invested in that workflow
Gemma 4 pricing
Gemma 4's weights are free to download and run under Google's Gemma license. Running it costs only whatever compute you use, whether local hardware or a cloud service like Vertex AI.
Prices in USD, checked on 27 September 2026. Current prices on ai.google.dev.
Main features
- Open weights
- Model weights are free to download and run under Google's Gemma license terms.
- Lightweight variants
- Sized to run on hardware ranging from cloud servers to laptops and phones.
- Multiple distribution channels
- Available through Hugging Face, Kaggle, Google AI Studio and Vertex AI.
- Runs on local tools
- Compatible with local runners such as Ollama for self-hosted use.
About Gemma 4
Gemma 4 is Google DeepMind's latest generation of openly available language models, following the earlier Gemma 1 through 3 releases. It was announced in April 2026, with additional variants and optimization techniques following through May and June 2026.
The Gemma line is built to be lightweight enough to run outside large data centers, from cloud servers down to laptops and phones, aimed at developers who want an open model they can run themselves rather than calling a hosted API.
Gemma models are distributed through Hugging Face, Kaggle and Google's own developer tooling, and can also be accessed through Google AI Studio and Vertex AI for those who want a managed environment instead of self-hosting.
- Made by
- Google DeepMind
- Launched
- 2026
- Free plan
- Yes, weights are free to download and run
- Available on
- Hugging Face, Kaggle, Self-hosted, Google AI Studio, Vertex AI
Usage and activity
Recent releases
Gemma 4 alternatives
Gemma 4 compared
Questions about Gemma 4
Is Gemma 4 free?
Yes, the model weights are free to download and run under Google's Gemma license, though that license includes Google's own usage terms rather than being a fully unrestricted open source license.
When was Gemma 4 released?
Google DeepMind announced Gemma 4 in April 2026, with further variants following through May and June 2026.
Where can I get Gemma 4?
It's distributed through Hugging Face, Kaggle and Google's own developer tools, and is also accessible through Google AI Studio and Vertex AI.
Can I run Gemma 4 on my laptop?
Yes, Gemma is designed to be lightweight enough to run on a range of hardware including laptops and phones, depending on the model size chosen.
Sources and links
Work on Gemma 4? Claim this listing. Something wrong or out of date? Suggest a correction. Stats come from the public GitHub and npm APIs and update every day.