Google: Gemma 4 26B A4B

google/gemma-4-26b-a4b-it

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...

Prompt $/1M

$0.070

Completion $/1M

$0.340

Context

262,144 tok

Best provider

52 tok/s · Wafer

Specs

Model id: google/gemma-4-26b-a4b-it
Display name: Google: Gemma 4 26B A4B
Modality: text+image+video->text
Tokenizer: Gemma
Instruct type: —
Context length: 262,144
Max completion: 16,384
Prompt $/1M: $0.070
Completion $/1M: $0.340
Free tier: no

Capabilities

Function callingno
JSON modeno
System promptno
Visionno
Moderatedno

Benchmark scores

Each axis is one benchmark, normalized to the best model in the field.

No benchmark data for this model.

Speed by provider

Provider	tok/s	TTFT	Total latency
Waferfastest	52	788 ms	903 ms
Ionstreamfastest	50	360 ms	454 ms
Artificial Analysis	45	682 ms	—
Google	34	487 ms	694 ms
Parasail	32	650 ms	797 ms
Cloudflare	29	1,880 ms	2,720 ms
Venice	28	1,051 ms	1,299 ms
Ambient	27	698 ms	1,022 ms
Novita	23	1,143 ms	2,044 ms
SiliconFlow	21	2,024 ms	5,671 ms
DeepInfra	21	402 ms	491 ms
Io Net	20	905 ms	1,639 ms
DekaLLM	18	3,710 ms	5,467 ms
NextBit	11	1,492 ms	1,897 ms