llmboard.aiAI model intelligence
Home

Model Rankings

OverallOpen ModelsAgentCodingReasoningMathKnowledgeInstruction FollowingTextVision
Image GenerationImage Editing
Video GenerationImage to VideoVideo Editing
Text to SpeechSpeech to Text
Embeddings

Efficiency

Chat Token PricingImage PricingVideo PricingAudio Pricing
Chat Speed & LatencyProvider Reliability

Benchmarks

GPQAMMLU-ProAIME 2025SWE-Bench VerifiedMMLUHumanity's Last ExamLiveCodeBenchMATHHumanEvalMMMU-Pro
All Benchmarks

Tools

Model Directory

Scoring & Data

Scoring & Data
1224 models729 benchmarks

Leaderboard Center

Overall RankingCodingCore BenchmarksPrice & ValueRuntime Performance

Modalities

All ModelsImage GenerationImage EditingVideo GenerationImage-to-VideoVideo EditingText-to-SpeechSpeech-to-TextEmbeddings

Data & Methods

Scoring MethodAll Benchmarks
llmboard.aiCopyright 2026 llmboard.ai

Google model product

Gemma 3 12B

Gemma 3 12B is a 12-billion-parameter vision-language model from Google that accepts text and image input and generates text output.

Updated Sep 8, 2026. Default version: Gemma 3 12B

LLMBoard Score10.0Gemma 3 12B
Coverage100%26 benchmark families
Context window131.1KTokens
Official input priceN/AOfficial price unavailable

On this page

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

Gemma 3 12B Capability Profile

This profile uses the model's current scored version. Arena ratings and prices are shown separately.

Gemma 3 12B LLMBoard score breakdown

Gemma 3 12B Benchmark Results

Benchmark scores for Gemma 3 12B.

28 rows
Columns

Show columns

Sort by
Benchmark
Score
Rank
Participants
Percentile
Evidence
Evaluated
BenchmarkVQAv2 (val)Score71.60%Rank01Participants3Percentile100.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkECLeKTicScore10.30%Rank03Participants8Percentile71.43%EvidenceCEvaluatedSep 8, 2026
BenchmarkBird-SQL (dev)Score47.90%Rank04Participants9Percentile62.50%EvidenceCEvaluatedSep 8, 2026
BenchmarkHiddenMathScore54.50%Rank04Participants13Percentile75.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkNatural2CodeScore80.70%Rank04Participants8Percentile57.14%EvidenceCEvaluatedSep 8, 2026
BenchmarkBIG-Bench HardScore85.70%Rank06Participants21Percentile75.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkFACTS GroundingScore75.80%Rank06Participants13Percentile58.33%EvidenceCEvaluatedSep 8, 2026
BenchmarkBIG-Bench Extra HardScore16.30%Rank08Participants11Percentile30.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkGlobal-MMLU-LiteScore69.50%Rank08Participants15Percentile50.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkInfoVQAScore64.90%Rank09Participants10Percentile11.11%EvidenceCEvaluatedSep 8, 2026
BenchmarkMMMU (val)Score59.60%Rank10Participants13Percentile25.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkMATHScore83.80%Rank14Participants71Percentile81.43%EvidenceCEvaluatedSep 8, 2026
BenchmarkTextVQAScore67.70%Rank14Participants16Percentile13.33%EvidenceCEvaluatedSep 8, 2026
BenchmarkWMT24++Score51.60%Rank15Participants23Percentile36.36%EvidenceCEvaluatedSep 8, 2026
BenchmarkGSM8kScore94.40%Rank16Participants48Percentile68.09%EvidenceCEvaluatedSep 8, 2026
BenchmarkMBPPScore73.00%Rank21Participants33Percentile37.50%EvidenceCEvaluatedSep 8, 2026
BenchmarkIFEvalScore88.90%Rank23Participants68Percentile67.16%EvidenceCEvaluatedSep 8, 2026
BenchmarkDocVQAScore87.10%Rank24Participants28Percentile14.81%EvidenceCEvaluatedSep 8, 2026
BenchmarkMathVista-MiniScore62.90%Rank24Participants25Percentile4.17%EvidenceCEvaluatedSep 8, 2026
BenchmarkAI2DScore84.20%Rank25Participants34Percentile27.27%EvidenceCEvaluatedSep 8, 2026
BenchmarkChartQAScore75.70%Rank25Participants26Percentile4.00%EvidenceCEvaluatedSep 8, 2026
BenchmarkHumanEvalScore85.40%Rank34Participants66Percentile49.23%EvidenceCEvaluatedSep 8, 2026
BenchmarkSimpleQAScore6.30%Rank43Participants47Percentile8.70%EvidenceCEvaluatedSep 8, 2026
BenchmarkLiveCodeBenchScore24.60%Rank67Participants75Percentile10.81%EvidenceCEvaluatedSep 8, 2026
BenchmarkMMLU-ProScore60.60%Rank111Participants138Percentile19.71%EvidenceCEvaluatedSep 8, 2026
BenchmarkLM Arena TextScore1,334.03 ratingRank145Participants210Percentile31.10%EvidenceAEvaluatedSep 2, 2026
BenchmarkLM Arena Text Style ControlScore1,341.68 ratingRank154Participants210Percentile26.79%EvidenceAEvaluatedSep 2, 2026
BenchmarkGPQAScore40.90%Rank216Participants247Percentile12.60%EvidenceCEvaluatedSep 8, 2026

Gemma 3 12B Arena Results

Preference and agent-evaluation results for the default version.

30 of 46 rows
Columns

Show columns

Sort by
Arena
Category
Rank
Rating / score
Votes
Observations
Result date
ArenatextCategorygermanRank104Rating / score1,370.25Votes166ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorygermanRank111Rating / score1,370.97Votes166ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry business and management and financial operationsRank112Rating / score1,376.02Votes364ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry legal and governmentRank117Rating / score1,380.95Votes217ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryindustry business and management and financial operationsRank120Rating / score1,385.69Votes364ObservationsN/AResult dateSep 2, 2026
ArenatextCategorycreative writingRank126Rating / score1,331.20Votes600ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryindustry legal and governmentRank131Rating / score1,382.26Votes217ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry life and physical and social scienceRank133Rating / score1,365.90Votes760ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorycreative writingRank133Rating / score1,332.62Votes600ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryrussianRank133Rating / score1,356.53Votes285ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryrussianRank135Rating / score1,334.42Votes285ObservationsN/AResult dateSep 2, 2026
ArenatextCategorynon englishRank140Rating / score1,317.65Votes1,499ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryindustry life and physical and social scienceRank141Rating / score1,368.18Votes760ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry mathematicalRank143Rating / score1,335.54Votes333ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorynon englishRank143Rating / score1,335.36Votes1,499ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry writing and literature and languageRank144Rating / score1,306.87Votes1,010ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryexclude tiesRank145Rating / score1,288.88Votes2,526ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry entertainment and sports and mediaRank145Rating / score1,297.43Votes697ObservationsN/AResult dateSep 2, 2026
ArenatextCategorymulti turnRank145Rating / score1,333.82Votes471ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryoverallRank145Rating / score1,334.03Votes3,829ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorymulti turnRank145Rating / score1,344.39Votes471ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryindustry mathematicalRank146Rating / score1,349.17Votes333ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorylonger queryRank146Rating / score1,345.95Votes371ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryindustry medicine and healthcareRank147Rating / score1,333.96Votes155ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryenglishRank148Rating / score1,348.48Votes2,330ObservationsN/AResult dateSep 2, 2026
ArenatextCategorylonger queryRank149Rating / score1,316.65Votes371ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategoryindustry entertainment and sports and mediaRank149Rating / score1,305.43Votes697ObservationsN/AResult dateSep 2, 2026
ArenatextCategorymathRank151Rating / score1,307.18Votes389ObservationsN/AResult dateSep 2, 2026
ArenatextCategoryinstruction followingRank153Rating / score1,299.25Votes1,145ObservationsN/AResult dateSep 2, 2026
Arenatext style controlCategorymathRank153Rating / score1,317.63Votes389ObservationsN/AResult dateSep 2, 2026

Gemma 3 12B Pricing

Official vendor API pricing appears first, followed by individual provider offers.

Official API
N/A
Official provider
N/A
Lowest third-party
From $0.05 input, $0.15 output per 1M via OpenRouter
Tracked offerings
4
4 rows
Columns

Show columns

Sort by
Provider
Provider model ID
Region
Input / 1M
Output / 1M
Context
Updated
ProviderOpenRouterProvider model IDgoogle/gemma-3-12b-itRegionglobalInput / 1M$0.05Output / 1M$0.15Context131.1KUpdatedSep 8, 2026
ProviderHeliconeProvider model IDgemma-3-12b-itRegionglobalInput / 1M$0.05Output / 1M$0.10Context131.1KUpdatedSep 8, 2026
ProviderNovitaAIProvider model IDgoogle/gemma-3-12b-itRegionglobalInput / 1M$0.05Output / 1M$0.10Context131.1KUpdatedSep 8, 2026
ProviderNeonProvider model IDgemma-3-12bRegionglobalInput / 1M$0.15Output / 1M$0.50Context131.1KUpdatedSep 8, 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 12B Runtime Performance

Provider-specific output speed and catalog latency for Gemma 3 12B. Runtime does not affect the capability score.

1 row
Columns

Show columns

Sort by
Provider
Output Speed
Catalog Latency
Max Input
Max Output
Updated
ProviderDeepInfraOutput Speed33.00 tok/sCatalog Latency0.20 sMax Input131.1KMax Output131.1KUpdatedSep 8, 2026

Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.

Gemma 3 12B Specifications

Technical details for the model's default version.

Version
Gemma 3 12B
Released
Mar 12, 2025
Knowledge cutoff
Unknown
Parameters
12B
Context window
131.1K
Max output
131.1K
Inputs
image, text
Outputs
text
Open weights
Yes
License
Gemma

Gemma 3 12B Versions

Available versions of this model. The score column identifies the version used in the overall ranking.

1 row
Columns

Show columns

Sort by
Version
Released
LLMBoard
Parameters
Context
Max output
Open weights
License
VersionGemma 3 12BReleasedMar 12, 2025LLMBoard9.97Parameters12BContext131.1KMax output131.1KOpen weightsYesLicenseGemma

Models similar to Gemma 3 12B

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

#243-0.02
GO

Gemma 4 E2B

Google

9.95 LLMBoard

Details
#241+0.21
GO

Gemini 1.5 Flash

Google

10.18 LLMBoard

Details
#228+4.52
GO

Gemma 3 27B

Google

14.49 LLMBoard

Details
#267-8.09
GO

Gemini Diffusion

Google

1.88 LLMBoard

Details
#216+9.29
GO

Gemini 2.5 Flash Lite

Google

19.26 LLMBoard

Details
#281-9.97
GO

MedGemma 4B IT

Google

0.00 LLMBoard

Details

What is Gemma 3 12B?

Key information about Gemma 3 12B and its available data.

Gemma 3 12B is a 12-billion-parameter vision-language model from Google with a 128K context window, multilingual support, and open weights. It is used for question answering, summarization, reasoning, and image understanding.

Data as of 2026-09-08.

FAQ

Common questions about Gemma 3 12B.

When was Gemma 3 12B released?

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

How much does Gemma 3 12B cost?

No official standard PAYG price is currently available for Gemma 3 12B. The lowest tracked third-party offer starts at $0.05 input and $0.15 output via OpenRouter.

Who created Gemma 3 12B?

Gemma 3 12B was created by Google.

What is the context window for Gemma 3 12B?

The default version has a 131.1K token context window.

Is Gemma 3 12B open weight?

Yes. The default version is marked as open weight under Gemma.

How many API providers offer Gemma 3 12B?

4 provider offerings are linked to the default version.