Gemma 4 26B A4B Instruct

Google llm

Sparse mixture-of-experts variant of Gemma 4 with 128 routed experts, built for efficient multimodal agentic workloads.

overall ⌄

52.3

36 of 93 ranked

price · $/M tokens

self-hosted

open weights, no first-party API

to run it yourself ⌄

16 GB

minimum · 62 GB recommended

Benchmarks

What its maker published, and what anyone else measured. Every figure links the document it came from.

benchmark	published	measured
MMLU-Pro	82.6	—
GPQA	82.3	—

No recorded lineage — root or standalone model.