llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsGemma 3 4B

Google model product

Gemma 3 4B

Gemma 3 4B is a 4-billion-parameter vision-language model from Google, handling text and image input and generating text output. It features a 128K context window, multilingual support, and open weights. Suitable for question answering, summarization, reasoning, and image understanding tasks.

Updated Aug 10, 2026. Default version: Gemma 3 4B

Compare
LLMBoard score22.3Gemma 3 4B
Coverage80%15 benchmark families
Context window131.1KTokens
Official input priceN/AOfficial price unavailable

On this page

  • Specification
  • Capability
  • Benchmarks
  • Arena
  • Pricing
  • Versions
  • About
  • Compare
  • Similar models
  • FAQ

Model specification

Structured fields from the published default version.

Version
Gemma 3 4B
Released
Mar 12, 2025
Knowledge cutoff
Aug 1, 2024
Parameters
4B
Context window
131.1K
Max output
131.1K
Inputs
image, text
Outputs
text
Open weights
No
License
Gemma

Capability profile

This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.

Gemma 3 4B category scores

Benchmark results

Published benchmark records for the scored version Gemma 3 4B.

26 rows
Columns

Show columns

VQAv2 (val)0.6330.0%CAug 7, 2026
ECLeKTic0.04857.1%CAug 7, 2026
Bird-SQL (dev)0.46716.7%CAug 7, 2026
HiddenMath0.471350.0%CAug 7, 2026
Natural2Code0.77814.3%CAug 7, 2026
FACTS Grounding0.791333.3%CAug 7, 2026
InfoVQA0.5990.0%CAug 7, 2026
BIG-Bench Extra Hard0.1101110.0%CAug 7, 2026
BIG-Bench Hard0.7112150.0%CAug 7, 2026
MMMU (val)0.511110.0%CAug 7, 2026
Global-MMLU-Lite0.513147.7%CAug 7, 2026
IFEval0.9156578.1%CAug 7, 2026
TextVQA0.615150.0%CAug 7, 2026
WMT24++0.5182322.7%CAug 7, 2026
MathVista-Mini0.523230.0%CAug 7, 2026
ChartQA0.724240.0%CAug 7, 2026
DocVQA0.826260.0%CAug 7, 2026
MATH0.8277162.9%CAug 7, 2026
GSM8k0.9284842.5%CAug 7, 2026
MBPP0.6303719.4%CAug 7, 2026
AI2D0.731323.2%CAug 7, 2026
SimpleQA0.043466.7%CAug 7, 2026
HumanEval0.7637617.3%CAug 7, 2026
LiveCodeBench0.172731.4%CAug 7, 2026
MMLU-Pro0.41221295.5%CAug 7, 2026
GPQA0.32202335.6%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

46 rows
Columns

Show columns

textgerman1841280.6162N/AAug 6, 2026
textindustry medicine and healthcare1891341.6197N/AAug 6, 2026
textindustry business and management and financial operations1991310.8390N/AAug 6, 2026
textindustry legal and government2011327.7245N/AAug 6, 2026
text style controlindustry medicine and healthcare2031351.2197N/AAug 6, 2026
textrussian2061294.2307N/AAug 6, 2026
text style controlgerman2071289.5162N/AAug 6, 2026
text style controlindustry business and management and financial operations2081329.2390N/AAug 6, 2026
text style controlrussian2111314.5307N/AAug 6, 2026
textindustry writing and literature and language2171288.21,080N/AAug 6, 2026
text style controlindustry legal and government2171339.2245N/AAug 6, 2026
textindustry life and physical and social science2191305.6780N/AAug 6, 2026
textcreative writing2201271.0666N/AAug 6, 2026
textnon english2201273.91,599N/AAug 6, 2026
textoverall2261290.94,171N/AAug 6, 2026
textexclude ties2291221.42,830N/AAug 6, 2026
text style controlindustry writing and literature and language2361297.81,080N/AAug 6, 2026
textenglish2381300.12,572N/AAug 6, 2026
text style controlnon english2391292.31,599N/AAug 6, 2026
textlonger query2401273.6444N/AAug 6, 2026
textexpert2431222.5208N/AAug 6, 2026
text style controlcreative writing2441275.3666N/AAug 6, 2026
text style controllonger query2461311.0444N/AAug 6, 2026
textindustry mathematical2471255.6365N/AAug 6, 2026
textindustry entertainment and sports and media2481233.5761N/AAug 6, 2026
texthard prompts2531252.81,092N/AAug 6, 2026
text style controlindustry mathematical2531274.3365N/AAug 6, 2026
textindustry software and it services2551269.5986N/AAug 6, 2026
text style controloverall2551303.44,171N/AAug 6, 2026
textinstruction following2561239.51,233N/AAug 6, 2026
textmulti turn2561255.1565N/AAug 6, 2026
text style controlexclude ties2571235.02,830N/AAug 6, 2026
text style controlindustry life and physical and social science2571314.9780N/AAug 6, 2026
textmath2581239.0423N/AAug 6, 2026
texthard prompts english2611253.5764N/AAug 6, 2026
text style controlenglish2641308.62,572N/AAug 6, 2026
text style controlexpert2651257.8208N/AAug 6, 2026
text style controlmath2651253.8423N/AAug 6, 2026
text style controlinstruction following2671267.51,233N/AAug 6, 2026
text style controlmulti turn2671271.9565N/AAug 6, 2026
text style controlhard prompts2691284.01,092N/AAug 6, 2026
textcoding2701230.8605N/AAug 6, 2026
text style controlindustry entertainment and sports and media2711248.6761N/AAug 6, 2026
text style controlindustry software and it services2731300.9986N/AAug 6, 2026
text style controlhard prompts english2781283.8764N/AAug 6, 2026
text style controlcoding2961273.9605N/AAug 6, 2026

Pricing

Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.

Official API
N/A
Official provider
N/A
Lowest third-party
From $0.05 input, $0.10 output per 1M via OpenRouter
Tracked offerings
1
1 rows
Columns

Show columns

OpenRoutergoogle/gemma-3-4b-itglobal$0.05$0.10131.1KAug 7, 2026

Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.

Gemma 3 4B versions

All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.

1 rows
Columns

Show columns

Gemma 3 4BMar 12, 202522.34B131.1K131.1KNoGemma

What is Gemma 3 4B?

A concise description based on the published model registry.

Gemma 3 4B is a 4-billion-parameter vision-language model from Google, handling text and image input and generating text output. It features a 128K context window, multilingual support, and open weights. Suitable for question answering, summarization, reasoning, and image understanding tasks.

Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.

Data snapshot: 2026-08-07. Editorial model content is not available in the backend.

Gemma 3 4B vs nearby models

Open a comparison with the three ranked models immediately above and below this model.

Gemma 3 4BvsMinistral 8BGemma 3 4BvsGPT-4.1-miniGemma 3 4BvsGemma 3n E4B LiteRTGemma 3 4BvsJamba 1.5 MiniGemma 3 4BvsIBM Granite 4.0 TinyGemma 3 4BvsQwen2.5 Omni 7B

Models similar to Gemma 3 4B

Recommendations prioritize the same model type and family, then the closest published LLMBoard score.

#150+0.2
GO

Gemma 3n E4B LiteRT

Google

22.5 LLMBoard

DetailsCompare
#157-4.4
GO

Gemma 3n E4B

Google

17.9 LLMBoard

DetailsCompare
#143+4.5
GO

Gemma 4 E4B

Google

26.8 LLMBoard

DetailsCompare
#142+4.9
GO

Gemini Diffusion

Google

27.2 LLMBoard

DetailsCompare
#136+7.9
GO

Gemma 2 9B

Google

30.2 LLMBoard

DetailsCompare
#161-8.0
GO

Gemma 4 E2B

Google

14.3 LLMBoard

DetailsCompare

FAQ

Common questions about Gemma 3 4B.

When was Gemma 3 4B released?

Gemma 3 4B's default version was released on Mar 12, 2025.

How much does Gemma 3 4B cost?

No official standard PAYG price is currently available for Gemma 3 4B. The lowest tracked third-party offer starts at $0.05 input and $0.10 output via OpenRouter.

Who created Gemma 3 4B?

Gemma 3 4B is published under Google in the model registry.

What is the context window for Gemma 3 4B?

The default version has a 131.1K token context window.

Is Gemma 3 4B open weight?

No. The default version is not marked as having publicly available weights.

How many API providers offer Gemma 3 4B?

1 published provider offerings are linked to the default version.

What models should I compare Gemma 3 4B with?

Nearby ranked alternatives include Ministral 8B, GPT-4.1-mini, Gemma 3n E4B LiteRT.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai