Gemma Models

gemma-4-31B-it

Gemma 4 31B is Google DeepMind’s dense 30.7B open model: text and image input, a 256K-token context window, and an Apache 2.0 license.

gemma-4-26B-A4B-it

Google’s Gemma 4 MoE: 25.2B parameters with 3.8B active, 256K context and image input, Apache 2.0. Runs at the speed of a model near 4B.

gemma-4-12B-it

Gemma 4 12B is Google DeepMind’s encoder-free multimodal model: text, image, audio and video input, a 256K context window, Apache 2.0.

diffusiongemma-26B-A4B-it

DiffusionGemma 26B A4B is Google DeepMind’s diffusion text model: a Gemma 4 MoE that denoises 256-token blocks in parallel, 15-20 tokens per pass.

gemma-4-E2B-it

Gemma 4 E2B is Google DeepMind’s smallest Gemma 4 model: 2.3B effective parameters, text, image and audio input, and a 128K context window.

gemma-3-270m

Gemma 3 270M is the smallest Gemma 3 model: 268M parameters, 32K context, trained on 6 trillion tokens. The Q8_0 GGUF is a 0.29 GB file.

gemma-3-1b-it

Gemma 3 1B is Google’s smallest Gemma 3 model: 1B parameters, a 32K context window and 2 trillion training tokens. The Q4_K_M GGUF build is 0.81 GB.