Google: Gemma 4 26B A4B

Text input Image input Video input Text output
Author's Description

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...

Key Specifications
Cost
$$
Context
262K
Parameters
26B
Released
Apr 03, 2026
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Temperature Include Reasoning Reasoning Presence Penalty Max Tokens Seed Structured Outputs Response Format Frequency Penalty Logit Bias Top P Stop
Features

This model supports the following features:

Structured Outputs Reasoning Response Format
Performance Summary

Google's Gemma 4 26B A4B IT demonstrates strong overall performance, particularly excelling in reliability with a perfect 100% success rate across all benchmarks, indicating consistent and usable responses. The model performs among the fastest models, typically ranking in the top tier for speed (73rd percentile), and offers competitive pricing, generally providing cost-effective solutions (76th percentile). In terms of specific benchmarks, Gemma 4 26B A4B IT shows excellent performance in mitigating hallucinations, achieving 98.0% accuracy by appropriately acknowledging uncertainty. Its instruction following capabilities are solid, with 71.0% accuracy, placing it in the 77th percentile for this category. Email classification also yielded strong results at 98.0% accuracy. While its duration for email classification was in the 51st percentile, its speed for hallucination and instruction following tests was notably high (82nd and 85th percentile, respectively). Key strengths include its exceptional reliability, strong hallucination mitigation, and efficient inference due to its MoE architecture, delivering near-31B quality at a fraction of the compute cost. No significant weaknesses were identified in the provided benchmarks.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.1
Completion $0.4
Input Cache Read $0.05

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Parasail
Parasail | google/gemma-4-26b-a4b-it-20260403 262K $0.1 / 1M tokens $0.4 / 1M tokens
Novita
Novita | google/gemma-4-26b-a4b-it-20260403 262K $0.13 / 1M tokens $0.4 / 1M tokens
Venice
Venice | google/gemma-4-26b-a4b-it-20260403 256K $0.08 / 1M tokens $0.35 / 1M tokens
Ionstream
Ionstream | google/gemma-4-26b-a4b-it-20260403 262K $0.08 / 1M tokens $0.4 / 1M tokens
Io Net
Io Net | google/gemma-4-26b-a4b-it-20260403 262K $0.08 / 1M tokens $0.35 / 1M tokens
NextBit
NextBit | google/gemma-4-26b-a4b-it-20260403 262K $0.13 / 1M tokens $0.4 / 1M tokens
DeepInfra
DeepInfra | google/gemma-4-26b-a4b-it-20260403 262K $0.08 / 1M tokens $0.35 / 1M tokens
DeepInfra
DeepInfra | google/gemma-4-26b-a4b-it-20260403 262K $0.08 / 1M tokens $0.35 / 1M tokens
Io Net
Io Net | google/gemma-4-26b-a4b-it-20260403 262K $0.09 / 1M tokens $0.4 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by google