Skip to content
Open release

Google releases Gemma 4 26B A4B

Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...

Type
OPEN_SOURCE_RELEASE
Detection
benchmark
Confidence
high
Status
Related entities
ModelGemma 4 26B A4B CompanyGoogle

Primary source

Epoch AI Benchmarking Hub · epoch.ai/benchmarks/eci ↗

Independent benchmark runs (GPQA Diamond, SWE-bench Verified, FrontierMath, ...) plus externally reported leaderboards and the Epoch Capabilities Index (ECI).

Related events