Spokes.wiki Search Graph Growth About

llm-providers-wiki

Software Application ↗ source url updated Fri Jun 05 2026 00:00:00 GMT+0000 (Coordinated Universal Time)

Gemma 4

google‘s open-weight (open-weight-models) model family — the locally-runnable, Apache-2.0 counterpart to its proprietary Gemini line. A multi-size lineup tuned for on-device / efficient deployment.

Family (2026)

VariantParamsNotes
E4Bsmallthe lightweight end
12B12B (dense)encoder-free multimodal (vision + native audio); runs on 16GB VRAM/unified memory; ~26B-level benchmarks at <½ the memory gemma-4-12b-announcement
26B-A4BMoE 25.2B total / 3.8B active, 256K ctxthe flagship; sparse activation for cheap inference open-source-llms-2026

Notable

Place in the market

Gemma 4 is google‘s entry in the open-weight wave alongside Meta Llama 4, Alibaba Qwen3, deepseek, Moonshot Kimi — competing on local-deploy efficiency and modality rather than raw frontier reasoning. The 12B’s encoder-free audio is the family’s current differentiator.

google · open-weight-models · quantization · gemma-4-qat · gemma-4-12b-announcement · open-source-llms-2026 · llm-benchmarks · deepseek