Google

Gemma 4 26B A4B

by Googletext+image+video->text11 endpoints
Neura Intelligence Index
53.7/ 100
Rank
#167
Confidence
medium
Benchmarks
5(13% coverage)

Overview

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...

Capabilities

Text generation
Image understanding
Video understanding
Tool use / Function calling
Structured output
Extended reasoning
Prompt caching
Long context
Top-P sampling
Stop sequences
Deterministic seed
Log probabilities

Modalities

Input
ImageTextVideo
Output
Text

Technical Specifications

Context Window
262.1K tokens
Max Output
235.9K tokens
Tokenizer
Gemma
Uptime (24h)
9996.9%
Parameters
25200000000B
License
apache_2_0Open Weights
Country
🇺🇸US
Release Date
April 2, 20265mo

Supported Parameters

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Benchmark Performance

general

BenchmarkScoreSource
MMMLU
86.3
verified
MRCR v2
44.1
verified

reasoning

BenchmarkScoreSource
GPQA Diamond
82.3
verified
Humanity's Last Exam
17.2
verified

multimodal

BenchmarkScoreSource
MMMU-Pro
73.8
verified

Performance Over Time

Pricing

Per 1M tokens
Input$0.09
Output$0.30
Cache read$0.05
Added
April 3, 2026
Last synced 9/19/2026