Skip to content

Gemma 4 26B A4B

ModelOpen sourceActive
by Google · family “gemmab-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... (description from the OpenRouter listing)

in: imagein: textin: videoout: textReasoningTool useStructured output
Epoch Capabilities Index
141.8 #101 of 268
90% CI 138.2 – 143.5
Listing ↗
Released
2 Apr 2026
Context window
262K tokens
Max output
236K tokens
Input $ / 1M tokens
$0.090
Output $ / 1M tokens
$0.30
Cached input $ / 1M
$0.050
Each benchmark shown separately with its own source

Benchmark results (6)

BenchmarkDomainScorevs best recordedSettingRunSource
LMCAagents29.7%
43%
——External ↗
DTBenchreasoning74.9%
76%
——External ↗
Chess Puzzlesgames6.0% ±2.4
8%
minimal6 Aug 2026Eval log ↗
WeirdMLcoding35.2% ±0.0
38%
——External ↗
OTIS Mock AIME 2024-2025math82.2% ±5.8
82%
minimal28 Aug 2026Eval log ↗
GPQA diamondscience73.2% ±3.2
76%
minimal6 Aug 2026Eval log ↗

Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.

API list price over time

Price history

$ per 1M tokens

7 recorded prices on 4 May 2026 (OpenRouter listing + Internet Archive snapshots); re-read on every data refresh, most recently 3h ago. Steps show when the price changed.

Events