llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsGemma 4 31B

Google model product

Gemma 4 31B

Gemma 4 31B is Google DeepMind's flagship dense multimodal model with 31 billion parameters and a 256K context window. Ranks #3 among open models on Arena AI. Built from the same research as Gemini 3, it features Per-Layer Embeddings, Shared KV Cache, alternating sliding-window and global attention, and variable aspect ratio vision encoding. Achieves an estimated LMArena text score of 1452.

Updated Aug 10, 2026. Default version: Gemma 4 31B

Compare
LLMBoard score74.3Gemma 4 31B
Coverage60%7 benchmark families
Context window262.1KTokens
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 4 31B
Released
Apr 2, 2026
Knowledge cutoff
Jan 1, 2025
Parameters
30.7B
Context window
262.1K
Max output
32.8K
Inputs
image, text
Outputs
text
Open weights
Yes
License
Apache 2.0

Capability profile

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

Gemma 4 31B category scores

Benchmark results

Published benchmark records for the scored version Gemma 4 31B.

12 rows
Columns

Show columns

BIG-Bench Extra Hard0.7111100.0%CAug 7, 2026
MedXpertQA0.651263.6%CAug 7, 2026
MRCR v2 (8-needle)0.752180.0%CAug 7, 2026
t2-bench0.952381.8%CAug 7, 2026
MathVision0.993274.2%CAug 7, 2026
AIME 20260.9111737.5%CAug 7, 2026
MMLU-Pro0.91512989.1%CAug 7, 2026
MMMLU0.9184964.6%CAug 7, 2026
LiveCodeBench v60.8215361.5%CAug 7, 2026
MMMU-Pro0.8246564.1%CAug 7, 2026
Humanity's Last Exam0.3479249.5%CAug 7, 2026
GPQA0.85923375.0%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

94 rows
Columns

Show columns

visioncaptioning181208.275N/AAug 6, 2026
visioncreative writing vision231278.41,904N/AAug 6, 2026
visionhomework241306.25,412N/AAug 6, 2026
vision style controlhomework241300.45,412N/AAug 6, 2026
vision style controlcaptioning251165.575N/AAug 6, 2026
text style controlspanish261466.1222N/AAug 6, 2026
visionentity recognition261219.1288N/AAug 6, 2026
visionchinese271331.52,050N/AAug 6, 2026
vision style controlchinese271311.52,050N/AAug 6, 2026
vision style controlcreative writing vision271259.11,904N/AAug 6, 2026
vision style controlentity recognition271190.8288N/AAug 6, 2026
agent praise complaintoverall28-0.0N/A10.6KAug 6, 2026
document style controloverall281445.210,577N/AJul 30, 2026
textrussian281461.2644N/AAug 6, 2026
agent task outcome explicitoverall29-0.0N/A34.8KAug 6, 2026
documentoverall291423.710,577N/AJul 30, 2026
visionhumor291270.91,253N/AAug 6, 2026
vision style controlhumor301250.81,253N/AAug 6, 2026
visionoverall311270.132,461N/AAug 6, 2026
textspanish321457.2222N/AAug 6, 2026
visionocr321277.622,914N/AAug 6, 2026
visiondiagram331281.28,375N/AAug 6, 2026
textmath341466.0398N/AAug 6, 2026
visionenglish341268.513,042N/AAug 6, 2026
vision style controlenglish341254.013,042N/AAug 6, 2026
vision style controloverall351255.732,461N/AAug 6, 2026
vision style controlocr351269.522,914N/AAug 6, 2026
vision style controldiagram361281.48,375N/AAug 6, 2026
text style controlmath371471.3398N/AAug 6, 2026
text style controlrussian381468.0644N/AAug 6, 2026
text style controlindustry mathematical391474.1319N/AAug 6, 2026
agent steerabilityoverall40-0.1N/A42.8KAug 6, 2026
textindustry mathematical401465.6319N/AAug 6, 2026
text factualityhard prompts english441484.91,522N/AAug 6, 2026
text style controlinstruction following451452.11,660N/AAug 6, 2026
agentoverall46-0.2N/A707.6KAug 6, 2026
agent bash recovery stepsoverall46-0.4N/A27.1KAug 6, 2026
agent tool hallucinationoverall46-0.3N/A592.3KAug 6, 2026
text factualityindustry software and it services471487.12,099N/AAug 6, 2026
text style controllonger query481466.81,643N/AAug 6, 2026
text factualitylonger query491463.91,643N/AAug 6, 2026
text style controlexpert501482.8446N/AAug 6, 2026
textindustry writing and literature and language511429.41,338N/AAug 6, 2026
textmulti turn511451.01,062N/AAug 6, 2026
text factualityinstruction following511445.51,659N/AAug 6, 2026
text factualityexclude ties521450.84,005N/AAug 6, 2026
text factualitynon english521437.63,262N/AAug 6, 2026
text style controlhard prompts english521482.71,523N/AAug 6, 2026
textexclude ties531439.64,008N/AAug 6, 2026
textinstruction following531431.71,660N/AAug 6, 2026
textnon english531431.53,263N/AAug 6, 2026
text factualityoverall531449.05,836N/AAug 6, 2026
text style controlindustry software and it services531490.52,100N/AAug 6, 2026
textlonger query541443.11,643N/AAug 6, 2026
text factualityindustry writing and literature and language541433.01,337N/AAug 6, 2026
textoverall551441.55,839N/AAug 6, 2026
textcreative writing551416.4926N/AAug 6, 2026
textindustry life and physical and social science551455.1877N/AAug 6, 2026
text factualitycoding551497.91,353N/AAug 6, 2026
text factualityhard prompts551470.83,351N/AAug 6, 2026
text style controlindustry life and physical and social science551473.3877N/AAug 6, 2026
text factualityenglish561458.42,574N/AAug 6, 2026
texthard prompts english581453.91,523N/AAug 6, 2026
text style controloverall581450.95,839N/AAug 6, 2026
text style controlexclude ties581453.84,008N/AAug 6, 2026
text style controlmulti turn581464.21,062N/AAug 6, 2026
textindustry software and it services591461.12,100N/AAug 6, 2026
text style controlindustry writing and literature and language591433.21,338N/AAug 6, 2026
textenglish601448.12,576N/AAug 6, 2026
textexpert601451.6446N/AAug 6, 2026
text style controlhard prompts601473.23,352N/AAug 6, 2026
text style controlnon english601438.83,263N/AAug 6, 2026
text factualitycreative writing611418.1925N/AAug 6, 2026
text style controlcoding611498.11,354N/AAug 6, 2026
text style controlenglish631459.82,576N/AAug 6, 2026
texthard prompts641445.53,352N/AAug 6, 2026
text style controlindustry legal and government641458.5431N/AAug 6, 2026
textindustry legal and government651442.4431N/AAug 6, 2026
text style controlindustry entertainment and sports and media661416.01,031N/AAug 6, 2026
text style controlcreative writing671420.6926N/AAug 6, 2026
webdevwebdev-react691356.36,470N/AAug 6, 2026
textindustry business and management and financial operations701426.81,129N/AAug 6, 2026
textindustry entertainment and sports and media701403.81,031N/AAug 6, 2026
textindustry medicine and healthcare711446.0401N/AAug 6, 2026
text style controlfrench711461.5239N/AAug 6, 2026
text style controlindustry business and management and financial operations711444.61,129N/AAug 6, 2026
text style controlindustry medicine and healthcare711465.3401N/AAug 6, 2026
textcoding721455.21,354N/AAug 6, 2026
webdevwebdev-html721373.9911N/AAug 6, 2026
textchinese741470.4372N/AAug 6, 2026
webdevoverall741364.07,772N/AAug 6, 2026
webdevwebdev741364.07,772N/AAug 6, 2026
text style controlchinese831476.6372N/AAug 6, 2026
textfrench841437.7239N/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 $0.10 input, $0.30 output per 1M via CrofAI
Tracked offerings
9
8 rows
Columns

Show columns

Poegoogle/gemma-4-31bglobalN/AN/A262.1KAug 7, 2026
Kenarigemma-4-31b-itglobalN/AN/A262.1KAug 7, 2026
CrofAIgemma-4-31b-itglobal$0.10$0.30262.1KAug 7, 2026
LLM Gatewaygemma-4-31b-itglobal$0.102$0.297262.1KAug 7, 2026
Neuralwattgemma-4-31bglobal$0.144$0.42262.1KAug 7, 2026
DigitalOceangemma-4-31B-itglobal$0.18$0.50256KAug 7, 2026
Cortecsgemma-4-31b-itglobal$0.223$0.39262KAug 7, 2026
NanoGPTTEE/gemma4-31bglobal$0.45$1262.1KAug 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 4 31B 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 4 31BApr 2, 202674.330.7B262.1K32.8KYesApache 2.0

What is Gemma 4 31B?

A concise description based on the published model registry.

Gemma 4 31B is Google DeepMind's flagship dense multimodal model with 31 billion parameters and a 256K context window. Ranks #3 among open models on Arena AI. Built from the same research as Gemini 3, it features Per-Layer Embeddings, Shared KV Cache, alternating sliding-window and global attention, and variable aspect ratio vision encoding. Achieves an estimated LMArena text score of 1452.

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 4 31B vs nearby models

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

Gemma 4 31BvsDeepSeek-V3.2Gemma 4 31BvsQwen3.5 397B A17BGemma 4 31BvsNova 2 ProGemma 4 31BvsClaude Sonnet 3.5Gemma 4 31BvsQwen3.7 PlusGemma 4 31BvsGemini 3 Flash

Models similar to Gemma 4 31B

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

#24-1.7
GO

Gemini 3 Flash

Google

72.6 LLMBoard

DetailsCompare
#14+3.7
GO

Gemini 2.5 Pro

Google

78.0 LLMBoard

DetailsCompare
#36-7.1
GO

Gemma 4 26B A4B

Google

67.2 LLMBoard

DetailsCompare
#8+7.9
GO

Gemini 3 Pro

Google

82.2 LLMBoard

DetailsCompare
#5+10.0
GO

Gemini 3.1 Pro

Google

84.3 LLMBoard

DetailsCompare
#42-10.7
GO

Gemini 2.5 Flash

Google

63.5 LLMBoard

DetailsCompare

FAQ

Common questions about Gemma 4 31B.

When was Gemma 4 31B released?

Gemma 4 31B's default version was released on Apr 2, 2026.

How much does Gemma 4 31B cost?

No official standard PAYG price is currently available for Gemma 4 31B. The lowest tracked third-party offer starts at $0.10 input and $0.30 output via CrofAI.

Who created Gemma 4 31B?

Gemma 4 31B is published under Google in the model registry.

What is the context window for Gemma 4 31B?

The default version has a 262.1K token context window.

Is Gemma 4 31B open weight?

Yes. The default version is marked as open weight under Apache 2.0.

How many API providers offer Gemma 4 31B?

9 published provider offerings are linked to the default version.

What models should I compare Gemma 4 31B with?

Nearby ranked alternatives include DeepSeek-V3.2, Qwen3.5 397B A17B, Nova 2 Pro.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai