Google model product
Gemma 3 4B is a 4-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 4B
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Gemma 3 4B.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkVQAv2 (val) | Score62.40% | Rank03 | Participants3 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkECLeKTic | Score4.60% | Rank04 | Participants8 | Percentile57.14% | EvidenceC | Evaluated |
| BenchmarkHiddenMath | Score43.00% | Rank07 | Participants13 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkNatural2Code | Score70.30% | Rank07 | Participants8 | Percentile14.29% | EvidenceC | Evaluated |
| BenchmarkBird-SQL (dev) | Score36.30% | Rank08 | Participants9 | Percentile12.50% | EvidenceC | Evaluated |
| BenchmarkFACTS Grounding | Score70.10% | Rank09 | Participants13 | Percentile33.33% | EvidenceC | Evaluated |
| BenchmarkBIG-Bench Extra Hard | Score11.00% | Rank10 | Participants11 | Percentile10.00% | EvidenceC | Evaluated |
| BenchmarkInfoVQA | Score50.00% | Rank10 | Participants10 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkBIG-Bench Hard | Score72.20% | Rank11 | Participants21 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkMMMU (val) | Score48.80% | Rank11 | Participants13 | Percentile16.67% | EvidenceC | Evaluated |
| BenchmarkGlobal-MMLU-Lite | Score54.50% | Rank14 | Participants15 | Percentile7.14% | EvidenceC | Evaluated |
| BenchmarkIFEval | Score90.20% | Rank15 | Participants68 | Percentile79.10% | EvidenceC | Evaluated |
| BenchmarkTextVQA | Score57.80% | Rank16 | Participants16 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkWMT24++ | Score46.80% | Rank18 | Participants23 | Percentile22.73% | EvidenceC | Evaluated |
| BenchmarkMathVista-Mini | Score50.00% | Rank25 | Participants25 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkChartQA | Score68.80% | Rank26 | Participants26 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMATH | Score75.60% | Rank27 | Participants71 | Percentile62.86% | EvidenceC | Evaluated |
| BenchmarkDocVQA | Score75.80% | Rank28 | Participants28 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkGSM8k | Score89.20% | Rank28 | Participants48 | Percentile42.55% | EvidenceC | Evaluated |
| BenchmarkMBPP | Score63.20% | Rank28 | Participants33 | Percentile15.63% | EvidenceC | Evaluated |
| BenchmarkAI2D | Score74.80% | Rank33 | Participants34 | Percentile3.03% | EvidenceC | Evaluated |
| BenchmarkSimpleQA | Score4.00% | Rank44 | Participants47 | Percentile6.52% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score71.30% | Rank56 | Participants66 | Percentile15.38% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench | Score12.60% | Rank74 | Participants75 | Percentile1.35% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score43.60% | Rank130 | Participants138 | Percentile5.84% | EvidenceC | Evaluated |
| BenchmarkLM Arena Text | Score1,290.61 rating | Rank165 | Participants210 | Percentile21.53% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text Style Control | Score1,303.12 rating | Rank180 | Participants210 | Percentile14.35% | EvidenceA | Evaluated |
| BenchmarkGPQA | Score30.80% | Rank234 | Participants247 | Percentile5.28% | EvidenceC | Evaluated |
Preference and agent-evaluation results for the default version.
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext | Categorygerman | Rank141 | Rating / score1,280.51 | Votes162 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry medicine and healthcare | Rank144 | Rating / score1,341.46 | Votes197 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry business and management and financial operations | Rank149 | Rating / score1,310.55 | Votes390 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry legal and government | Rank151 | Rating / score1,327.32 | Votes245 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry medicine and healthcare | Rank154 | Rating / score1,351.94 | Votes197 | ObservationsN/A | Result date |
| Arenatext | Categoryrussian | Rank155 | Rating / score1,293.04 | Votes307 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry business and management and financial operations | Rank155 | Rating / score1,329.16 | Votes390 | ObservationsN/A | Result date |
| Arenatext style control | Categorygerman | Rank156 | Rating / score1,289.53 | Votes162 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry writing and literature and language | Rank160 | Rating / score1,287.88 | Votes1,080 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry life and physical and social science | Rank161 | Rating / score1,305.53 | Votes780 | ObservationsN/A | Result date |
| Arenatext style control | Categoryrussian | Rank161 | Rating / score1,313.22 | Votes307 | ObservationsN/A | Result date |
| Arenatext | Categorycreative writing | Rank163 | Rating / score1,270.34 | Votes666 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry legal and government | Rank163 | Rating / score1,339.29 | Votes245 | ObservationsN/A | Result date |
| Arenatext | Categorynon english | Rank165 | Rating / score1,272.92 | Votes1,599 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank165 | Rating / score1,290.61 | Votes4,171 | ObservationsN/A | Result date |
| Arenatext | Categoryexclude ties | Rank166 | Rating / score1,220.80 | Votes2,830 | ObservationsN/A | Result date |
| Arenatext | Categoryenglish | Rank173 | Rating / score1,300.38 | Votes2,572 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry writing and literature and language | Rank173 | Rating / score1,297.46 | Votes1,080 | ObservationsN/A | Result date |
| Arenatext | Categorylonger query | Rank174 | Rating / score1,273.25 | Votes444 | ObservationsN/A | Result date |
| Arenatext style control | Categorynon english | Rank174 | Rating / score1,291.08 | Votes1,599 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry mathematical | Rank175 | Rating / score1,255.66 | Votes365 | ObservationsN/A | Result date |
| Arenatext style control | Categorycreative writing | Rank177 | Rating / score1,274.59 | Votes666 | ObservationsN/A | Result date |
| Arenatext style control | Categorylonger query | Rank177 | Rating / score1,311.18 | Votes444 | ObservationsN/A | Result date |
| Arenatext | Categoryexpert | Rank179 | Rating / score1,222.98 | Votes208 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry mathematical | Rank180 | Rating / score1,274.32 | Votes365 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank180 | Rating / score1,303.12 | Votes4,171 | ObservationsN/A | Result date |
| Arenatext | Categoryhard prompts | Rank181 | Rating / score1,252.40 | Votes1,092 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry entertainment and sports and media | Rank181 | Rating / score1,232.70 | Votes761 | ObservationsN/A | Result date |
| Arenatext | Categorymath | Rank181 | Rating / score1,238.98 | Votes423 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry software and it services | Rank182 | Rating / score1,268.94 | Votes986 | ObservationsN/A | Result date |
Official vendor API pricing appears first, followed by individual provider offers.
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderOpenRouter | Provider model IDgoogle/gemma-3-4b-it | Regionglobal | Input / 1M$0.05 | Output / 1M$0.10 | Context131.1K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Provider-specific output speed and catalog latency for Gemma 3 4B. Runtime does not affect the capability score.
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderDeepInfra | Output Speed33.00 tok/s | Catalog Latency0.20 s | Max Input131.1K | Max Output131.1K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Technical details for the model's default version.
Available versions of this model. The score column identifies the version used in the overall ranking.
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionGemma 3 4B | Released | LLMBoard0.00 | Parameters4B | Context131.1K | Max output131.1K | Open weightsYes | LicenseGemma |
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Gemma 3 4B and its available data.
Gemma 3 4B is a Google vision-language model with a 128K context window, multilingual support, and open weights. It is designed for question answering, summarization, reasoning, and image understanding tasks.
Data as of 2026-09-08.
Common questions about Gemma 3 4B.
Gemma 3 4B's default version was released on Mar 12, 2025.
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.
Gemma 3 4B was created by Google.
The default version has a 131.1K token context window.
Yes. The default version is marked as open weight under Gemma.
1 provider offerings are linked to the default version.