Gemma 4 26B A4B

Gemma 4 26B A4B is a reasoning model from Google in the Gemma 4 family. 17 benchmarks count toward its index, in 8 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-04
RankedEnough results for an index.
IndexAll categories, out of 100.44.5±3.9Median 51.0
CoverageIndex weight that has results.100%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.FreeMedian $0.60
Output / 1MUS dollars per 1M output tokens.FreeMedian $2.55
ContextMaximum tokens in one request.256KMedian 262K
EloLMArena rating and rank.1434±7 #92Median 1422

Run locally

25.81B parameters
16-bit
55.2 GB
8-bit
29.8 GB
4-bit
17.4 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 24 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.2 counted34.0190 / 218
CodingCode writing and repair.3 counted45.3161 / 209
ReasoningLogic problems and puzzles.3 counted42.1153 / 227
MultimodalTasks with images and text.3 counted49.668 / 84
KnowledgeFacts and expert knowledge.3 counted46.7161 / 268
MultilingualTasks in many languages.1 counted36.3—
InstructionTasks with strict prompt rules.1 counted56.7—
MathMath problems.1 counted56.8—

Results

17 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
EuroEval PortugueseMultilingual62.0% ±0.524 / 182
EuroEval FrenchMultilingual65.2% ±0.525 / 178
EuroEval ItalianMultilingual61.1% ±0.526 / 174
EuroEval PolishMultilingual55.5% ±0.739 / 209
EuroEval SwedishMultilingual58.8% ±0.637 / 195
EuroEval DutchMultilingual56.4% ±0.538 / 175
EuroEval GermanMultilingual49.1% ±0.643 / 191
Artificial Analysis IFBenchInstruction72.4%36 / 146
EuroEval SpanishMultilingual49.4% ±0.545 / 176
OTIS Mock AIME 2024-2025Math82.2% ±5.881 / 200
EuroEval SwedishMultilingual44.4% ±0.985 / 195
Graduate-Level Google-Proof Q&AKnowledge82.3%62 / 141
EuroEval DutchMultilingual45.3% ±0.783 / 175
EuroEval GermanMultilingual37.7% ±1.092 / 191
GPQA diamond · Epoch AIKnowledge73.2% ±3.2108 / 217
EuroEval PortugueseMultilingual43.5% ±0.993 / 182
Artificial Analysis GPQA DiamondKnowledge79.2%102 / 199
EuroEval SpanishMultilingual35.4% ±0.693 / 176
EuroEval ItalianMultilingual41.0% ±0.994 / 174
LiveCodeBench v6Coding77.1%33 / 59
EuroEval FrenchMultilingual44.4% ±1.0100 / 178
EuroEval PolishMultilingual38.5% ±0.9121 / 209
Artificial Analysis MMMU-ProMultimodal69.2%77 / 113
Critical Physics TasksReasoning0.0%150 / 214
GDPval-AA normalizedAgentic3.2%98 / 133
Artificial Analysis SciCodeCoding40.0%88 / 114
MRCRv2Reasoning44.1%11 / 14
Humanity's Last ExamKnowledge17.2%69 / 77

Sources

BenchLMCC BY-NC 4.0Epoch AICC BYEuroEvalMIT

History

5 releases
Index44.9 → 44.5
23 Oct 202612 Nov 2026
Output / 1MFree → Free
23 Oct 202612 Nov 2026

Questions

How much does Gemma 4 26B A4B cost?
Input tokens are free, output tokens are free.
How good is Gemma 4 26B A4B?
Its index is 44.5 out of 100. That is place 181 of 283 ranked models.
What is Gemma 4 26B A4B best at?
Knowledge: 46.7 out of 100, place 161 of 268.
What is the context window of Gemma 4 26B A4B?
256K tokens.
Is Gemma 4 26B A4B open source?
Yes. Its weights are published.
When was Gemma 4 26B A4B released?
April 2026.

Badge

Gemma 4 26B A4B on Inference360: place 181 of 283, index 44.5
HTML
<a href="https://inference360.ai/models/gemma-4-26b-a4b"><img src="https://inference360.ai/badges/gemma-4-26b-a4b.svg" alt="Gemma 4 26B A4B on Inference360: place 181 of 283, index 44.5" height="20"></a>
Markdown
[![Gemma 4 26B A4B on Inference360: place 181 of 283, index 44.5](https://inference360.ai/badges/gemma-4-26b-a4b.svg)](https://inference360.ai/models/gemma-4-26b-a4b)

Compare with

o144.5DeepSeek-R1-052844.6Grok 3 Mini44.3Qwen3.5-4B44.3Gemma 4 31B47.9

More from Google

All models
Gemma 4 31B47.9Gemma 4 E4B28.2Gemma 4 E2B21.9Lyria 3 Pro PreviewUnrankedLyria 3 Clip PreviewUnranked