Google · AI MODEL

Google: 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...

Input / 1M: $0.0675
Output / 1M: $0.225
Cached input / 1M: $0.0375

Specifications

Context window: 262,144
Input modalities: text, image
Output modalities: text
Last checked: 10/18/2026, 11:00:43 PM

Source

Prices come from the linked provider source and represent list prices before individual discounts.

Open source

Hugging Face

Inference offers

Compare the live Hugging Face routes available for this model.

Provider	Catalog	Status	Input / 1M	Output / 1M	Context window	Latency	Throughput	Features
Deepinfra	HUGGINGFACE	live	$0.070	$0.340	262,144	461.4 ms	91.9 tok/s	Tools, Structured output
Featherless Ai	HUGGINGFACE	live	—	—	—	—	—	—
Novita	HUGGINGFACE	live	$0.130	$0.400	262,144	750.4 ms	140 tok/s	Tools, Structured output
Scaleway	HUGGINGFACE	live	$0.285	$0.570	—	376.2 ms	192.4 tok/s	Tools, Structured output

History

Price history

Input / 1MOutput / 1M$0.340 max

Date	Input / 1M	Output / 1M	Confidence
10/16/2026, 8:00:51 PM	$0.0675	$0.225	100%
10/16/2026, 7:00:11 AM	$0.090	$0.300	100%
10/15/2026, 8:00:35 PM	$0.0765	$0.255	100%
10/12/2026, 5:00:59 AM	$0.042	$0.220	100%
9/10/2026, 5:00:19 PM	$0.042	$0.220	100%
9/7/2026, 7:57:41 AM	$0.070	$0.340	100%

Smart alternatives

Cheaper alternatives

ApodexAI MODEL

Apodex: Apodex 1.1 Mini (free)

Apodex 1.1 Mini is a reasoning-first model from Apodex, built for complex, long-horizon research and forecasting tasks. It works directly with files, data, code, and tools to produce verifiable results,...

Input / 1M: $0.000
Output / 1M: $0.000

CohereAI MODEL

Cohere: North Mini Code (free)

North Mini Code is Cohere's first agentic coding model and the debut of its North family. A sparse mixture-of-experts model with 30B total parameters and 3B active, it is optimized...

Input / 1M: $0.000
Output / 1M: $0.000

DeepSeekAI MODEL

DeepSeek: DeepSeek V4 Flash 0731 (free)

DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....

Input / 1M: $0.000
Output / 1M: $0.000

Dots StudioAI MODEL

Dots Studio: Dots3-Note Preview (free)

Dots3-Note Preview is an open-weight mixture-of-experts model from Dots Studio, with 16B active parameters out of 280B total. It is the lightest model in the Dots 3 family and is...