Google Unveils Gemma 4! The smartest open models yet, now fully open under the Apache 2.0 license

Features native function calling and JSON output for complex, multi-step problem solving.

[{"selector":"#anim-12780f33-6211-46eb-ac5b-816c8345831e","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] Brush Stroke

– The edge models don't just read text—they natively process images and audio in real-time

Red Section Separator

A Size for Every Device

[{"selector":"#anim-eda56f75-1270-4eb9-90b4-e49a761a7984 [data-leaf-element=\"true\"]","keyframes":{"transform":["translate3d(5.4597497563320845%, 0, 0) translate(0%, 0%) scale(1)","translate3d(0%, 0, 0) translate(0%, 0%) scale(1)"]},"delay":0,"duration":3000,"fill":"forwards"}] [{"selector":"#anim-669cd608-d616-4188-b924-79df2f43c17b","keyframes":{"opacity":[0,1]},"delay":300,"duration":1600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] From mobile-friendly E2B & E4B to the powerhouse 26B & 31B models for servers. Learn more

Accelerated by NVIDIA

[{"selector":"#anim-128d832f-60b2-40ac-8048-f2916db44db9","keyframes":{"opacity":[0,1]},"delay":0,"duration":2000,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}]

Join the Open AI Era

[{"selector":"#anim-09462384-3fcb-4d53-8552-2ecccc355deb","keyframes":{"opacity":[0,1]},"delay":0,"duration":1600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-d31ac4c7-d9a7-4843-acdf-db2334ef6287","keyframes":{"transform":["translate3d(0px, 76.94147%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1600,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-90d5fa55-3cf0-406a-b7b5-e992d7a602fe","keyframes":{"opacity":[0,1]},"delay":200,"duration":2000,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] With 400M+ Gemma downloads, the community is booming. Start building today! Learn more

Google Unveils Gemma 4! The smartest open models yet, now fully open under the Apache 2.0 license

Features native function calling and JSON output for complex, multi-step problem solving.

• Built for AI Agents

– The edge models don't just read text—they natively process images and audio in real-time

True Multimodal Power 👁️

A Size for Every Device

Accelerated by NVIDIA

Perfectly optimized for max performance on RTX PCs and Jetson edge devices.

Join the Open AI Era

Join the Open AI Era