Inference
EN
Google

Gemma 4 E4B

Gemma 4 E4B is a reasoning model from Google in the Gemma 4 family. 15 benchmarks count toward its index, in 7 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-04
RankedEnough results for an index.
IndexAll categories, out of 100.28.2±5.2Median 51.0
CoverageIndex weight that has results.90%
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.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.—Median 1422

Run locally

8B parameters
16-bit
17.8 GB
8-bit
9.9 GB
4-bit
6.1 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 8 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.2 counted25.3212 / 218
CodingCode writing and repair.2 counted27.4201 / 209
ReasoningLogic problems and puzzles.3 counted30.9206 / 227
MultimodalTasks with images and text.3 counted32.880 / 84
KnowledgeFacts and expert knowledge.3 counted30.2238 / 268
MultilingualTasks in many languages.1 counted8.8—
InstructionTasks with strict prompt rules.1 counted30.3—
MathMath problems.None counted——

Results

15 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Artificial Analysis IFBenchInstruction44.2%88 / 146
Critical Physics TasksReasoning0.6%132 / 214
EuroEval SwedishMultilingual29.2% ±1.2134 / 195
EuroEval DutchMultilingual33.2% ±0.8121 / 175
EuroEval GermanMultilingual21.7% ±0.9142 / 191
EuroEval SpanishMultilingual23.6% ±0.8131 / 176
Massive Multitask Language Understanding ProfessionalKnowledge69.4%56 / 75
EuroEval ItalianMultilingual26.4% ±0.9130 / 174
EuroEval FrenchMultilingual30.1% ±0.9135 / 178
GDPval-AA normalizedAgentic0.0%101 / 133
EuroEval PortugueseMultilingual27.6% ±0.9140 / 182
EuroEval PolishMultilingual25.2% ±1.0163 / 209
Graduate-Level Google-Proof Q&AKnowledge58.6%114 / 141
Artificial Analysis GPQA DiamondKnowledge57.6%162 / 199
Artificial Analysis Long Context ReasoningReasoning32.0%177 / 217
LiveCodeBench v6Coding52.0%51 / 59
τ²-Bench Tool-Agent-User EvaluationAgentic20.8%130 / 149
Artificial Analysis MMMU-ProMultimodal51.4%102 / 113
Massive Multi-discipline Multimodal Understanding ProMultimodal52.6%48 / 53
Artificial Analysis Humanity's Last ExamKnowledge3.8%202 / 222
ExtractBench: schema-guided enterprise document extraction (LlamaIndex leaderboard)Multimodal69.6%24 / 26
MRCRv2Reasoning25.4%13 / 14
ParseBench: document parsing for AI agents (LlamaIndex leaderboard)Multimodal41.1%37 / 39
Artificial Analysis SciCodeCoding24.4%109 / 114

Sources

BenchLMCC BY-NC 4.0EuroEvalMIT

History

5 releases
Index28.1 → 28.2
23 Oct 202612 Nov 2026

Questions

How good is Gemma 4 E4B?
Its index is 28.2 out of 100. That is place 255 of 283 ranked models.
What is Gemma 4 E4B best at?
Knowledge: 30.2 out of 100, place 238 of 268.
What is the context window of Gemma 4 E4B?
128K tokens.
Is Gemma 4 E4B open source?
Yes. Its weights are published.
When was Gemma 4 E4B released?
April 2026.

Badge

Gemma 4 E4B on Inference360: place 255 of 283, index 28.2
HTML
<a href="https://inference360.ai/models/gemma-4-e4b"><img src="https://inference360.ai/badges/gemma-4-e4b.svg" alt="Gemma 4 E4B on Inference360: place 255 of 283, index 28.2" height="20"></a>
Markdown
[![Gemma 4 E4B on Inference360: place 255 of 283, index 28.2](https://inference360.ai/badges/gemma-4-e4b.svg)](https://inference360.ai/models/gemma-4-e4b)

Compare with

Devstral Small28.2Llama 3.3 70B28.2Mistral Large 227.4Gemma 4 E2B21.9Gemma 4 26B A4B44.5

More from Google

All models
Gemma 4 31B47.9Gemma 4 26B A4B44.5Gemma 4 E2B21.9Lyria 3 Pro PreviewUnrankedLyria 3 Clip PreviewUnranked