Google model product
Gemma 4 31B is a Google dense multimodal model with 31 billion parameters and a 256K context window.
Updated Sep 8, 2026. Default version: Gemma 4 31B
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Gemma 4 31B.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkBIG-Bench Extra Hard | Score74.40% | Rank01 | Participants11 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkMedXpertQA | Score61.30% | Rank05 | Participants12 | Percentile63.64% | EvidenceC | Evaluated |
| Benchmarkt2-bench | Score86.40% | Rank05 | Participants23 | Percentile81.82% | EvidenceC | Evaluated |
| BenchmarkMRCR v2 (8-needle) | Score66.40% | Rank08 | Participants24 | Percentile69.57% | EvidenceC | Evaluated |
| BenchmarkMathVision | Score85.60% | Rank12 | Participants35 | Percentile67.65% | EvidenceC | Evaluated |
| BenchmarkAIME 2026 | Score89.20% | Rank16 | Participants22 | Percentile28.57% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score85.20% | Rank18 | Participants138 | Percentile87.59% | EvidenceC | Evaluated |
| BenchmarkMMMLU | Score88.40% | Rank18 | Participants49 | Percentile64.58% | EvidenceC | Evaluated |
| BenchmarkMMMU-Pro | Score76.90% | Rank24 | Participants69 | Percentile66.18% | EvidenceC | Evaluated |
| BenchmarkLM Arena Document Style Control | Score1,445.18 rating | Rank24 | Participants33 | Percentile28.13% | EvidenceA | Evaluated |
| BenchmarkLM Arena Document | Score1,423.70 rating | Rank25 | Participants33 | Percentile25.00% | EvidenceA | Evaluated |
| BenchmarkLM Arena Vision | Score1,275.94 rating | Rank25 | Participants107 | Percentile77.36% | EvidenceA | Evaluated |
| BenchmarkLiveCodeBench v6 | Score80.00% | Rank26 | Participants62 | Percentile59.02% | EvidenceC | Evaluated |
| BenchmarkLM Arena Vision Style Control | Score1,261.28 rating | Rank28 | Participants107 | Percentile74.53% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Praise Complaint | Score-4.53% | Rank34 | Participants49 | Percentile31.25% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Task Outcome Explicit | Score-3.41% | Rank35 | Participants49 | Percentile29.17% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text Factuality | Score1,449.02 rating | Rank45 | Participants121 | Percentile63.33% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text | Score1,441.82 rating | Rank46 | Participants210 | Percentile78.47% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Steerability | Score-10.70% | Rank48 | Participants49 | Percentile2.08% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Leaderboard | Score-23.70% | Rank49 | Participants49 | Percentile0.00% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Bash Recovery Steps | Score-71.50% | Rank49 | Participants49 | Percentile0.00% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Tool Hallucination | Score-28.37% | Rank49 | Participants49 | Percentile0.00% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text Style Control | Score1,451.28 rating | Rank50 | Participants210 | Percentile76.56% | EvidenceA | Evaluated |
| BenchmarkHumanity's Last Exam | Score26.50% | Rank56 | Participants103 | Percentile46.08% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score84.30% | Rank67 | Participants247 | Percentile73.17% | EvidenceC | Evaluated |
| BenchmarkLM Arena Webdev | Score1,363.07 rating | Rank69 | Participants97 | Percentile29.17% | EvidenceA | Evaluated |
Preference and agent-evaluation results for the default version.
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenavision | Categorycreative writing vision | Rank15 | Rating / score1,281.52 | Votes2,138 | ObservationsN/A | Result date |
| Arenavision | Categorycaptioning | Rank18 | Rating / score1,209.69 | Votes84 | ObservationsN/A | Result date |
| Arenavision style control | Categoryhomework | Rank18 | Rating / score1,302.34 | Votes5,633 | ObservationsN/A | Result date |
| Arenavision | Categorychinese | Rank19 | Rating / score1,330.71 | Votes2,181 | ObservationsN/A | Result date |
| Arenavision | Categoryhomework | Rank20 | Rating / score1,308.01 | Votes5,633 | ObservationsN/A | Result date |
| Arenavision style control | Categorycreative writing vision | Rank20 | Rating / score1,261.07 | Votes2,138 | ObservationsN/A | Result date |
| Arenavision style control | Categorychinese | Rank21 | Rating / score1,311.50 | Votes2,181 | ObservationsN/A | Result date |
| Arenadocument style control | Categoryoverall | Rank24 | Rating / score1,445.18 | Votes10,577 | ObservationsN/A | Result date |
| Arenavision | Categoryhumor | Rank24 | Rating / score1,273.20 | Votes1,366 | ObservationsN/A | Result date |
| Arenavision style control | Categorycaptioning | Rank24 | Rating / score1,165.26 | Votes84 | ObservationsN/A | Result date |
| Arenavision style control | Categoryentity recognition | Rank24 | Rating / score1,189.70 | Votes295 | ObservationsN/A | Result date |
| Arenadocument | Categoryoverall | Rank25 | Rating / score1,423.70 | Votes10,577 | ObservationsN/A | Result date |
| Arenavision | Categoryentity recognition | Rank25 | Rating / score1,218.32 | Votes295 | ObservationsN/A | Result date |
| Arenavision | Categoryoverall | Rank25 | Rating / score1,275.94 | Votes34,348 | ObservationsN/A | Result date |
| Arenatext | Categoryrussian | Rank26 | Rating / score1,460.52 | Votes762 | ObservationsN/A | Result date |
| Arenavision | Categoryocr | Rank26 | Rating / score1,281.67 | Votes24,110 | ObservationsN/A | Result date |
| Arenavision style control | Categoryhumor | Rank27 | Rating / score1,253.22 | Votes1,366 | ObservationsN/A | Result date |
| Arenavision style control | Categoryocr | Rank28 | Rating / score1,273.30 | Votes24,110 | ObservationsN/A | Result date |
| Arenavision style control | Categoryoverall | Rank28 | Rating / score1,261.28 | Votes34,348 | ObservationsN/A | Result date |
| Arenavision | Categorydiagram | Rank29 | Rating / score1,281.79 | Votes8,791 | ObservationsN/A | Result date |
| Arenavision | Categoryenglish | Rank30 | Rating / score1,269.44 | Votes13,778 | ObservationsN/A | Result date |
| Arenavision style control | Categorydiagram | Rank30 | Rating / score1,281.95 | Votes8,791 | ObservationsN/A | Result date |
| Arenatext | Categorymath | Rank31 | Rating / score1,466.50 | Votes402 | ObservationsN/A | Result date |
| Arenatext style control | Categorymath | Rank31 | Rating / score1,471.58 | Votes402 | ObservationsN/A | Result date |
| Arenavision style control | Categoryenglish | Rank31 | Rating / score1,254.80 | Votes13,778 | ObservationsN/A | Result date |
| Arenatext | Categoryspanish | Rank32 | Rating / score1,453.48 | Votes245 | ObservationsN/A | Result date |
| Arenatext style control | Categoryrussian | Rank33 | Rating / score1,468.04 | Votes762 | ObservationsN/A | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank34 | Rating / score-0.05 | VotesN/A | Observations10.7K | Result date |
| Arenatext | Categoryindustry mathematical | Rank34 | Rating / score1,465.48 | Votes325 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry mathematical | Rank34 | Rating / score1,473.17 | Votes325 | 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 |
|---|
| ProviderNvidia | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context256K | Updated |
| ProviderKenari | Provider model IDgemma-4-31b-it | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context262.1K | Updated |
| ProviderRequesty | Provider model IDgemma-4-31b-it | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context262.1K | Updated |
| ProviderPoe | Provider model IDgoogle/gemma-4-31b | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context262.1K | Updated |
| ProviderOpenRouter | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.09 | Output / 1M$0.34 | Context262.1K | Updated |
| ProviderKilo Gateway | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.09 | Output / 1M$0.34 | Context262.1K | Updated |
| ProviderNanoGPT | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.10 | Output / 1M$0.45 | Context262.1K | Updated |
| ProviderCrofAI | Provider model IDgemma-4-31b-it | Regionglobal | Input / 1M$0.10 | Output / 1M$0.30 | Context262.1K | Updated |
| ProviderWeights & Biases | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.10 | Output / 1M$0.34 | Context262.1K | Updated |
| ProviderDevPass (LLM Gateway) | Provider model IDgemma-4-31b-it | Regionglobal | Input / 1M$0.102 | Output / 1M$0.297 | Context262.1K | Updated |
| ProviderLilac | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.11 | Output / 1M$0.35 | Context262.1K | Updated |
| ProviderOrcaRouter | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.13 | Output / 1M$0.38 | Context262.1K | Updated |
| ProviderDeep Infra | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.13 | Output / 1M$0.38 | Context262.1K | Updated |
| ProviderNEAR AI Cloud | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.13 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderFastRouter | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.13 | Output / 1M$0.38 | Context262.1K | Updated |
| ProviderSiliconFlow | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.13 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderHugging Face | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderCrusoe | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderMerge Gateway | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderFriendli | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderVercel AI Gateway | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderNovitaAI | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderAbacus | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.14 | Output / 1M$0.40 | Context262.1K | Updated |
| ProviderNeuralwatt | Provider model IDgemma-4-31b | Regionglobal | Input / 1M$0.144 | Output / 1M$0.42 | Context262.1K | Updated |
| ProviderDigitalOcean | Provider model IDgemma-4-31B-it | Regionglobal | Input / 1M$0.18 | Output / 1M$0.50 | Context256K | Updated |
| Providerai& | Provider model IDgoogle/gemma-4-31b-it | Regionglobal | Input / 1M$0.20 | Output / 1M$0.50 | Context262.1K | Updated |
| ProviderCortecs | Provider model IDgemma-4-31b-it | Regionglobal | Input / 1M$0.223 | Output / 1M$0.39 | Context262K | Updated |
| ProviderInfomaniak | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.25 | Output / 1M$0.50 | Context100K | Updated |
| ProviderBerget.AI | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.275 | Output / 1M$0.55 | Context128K | Updated |
| ProviderTogether AI | Provider model IDgoogle/gemma-4-31B-it | Regionglobal | Input / 1M$0.39 | Output / 1M$0.97 | Context262.1K | Updated |
Provider-specific output speed and catalog latency for Gemma 4 31B. Runtime does not affect the capability score.
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderFriendliAI | Output Speed22.86 tok/s | Catalog Latency22.43 s | Max Input262.1K | 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 4 31B | Released | LLMBoard58.54 | Parameters30.7B | Context262.1K | Max output131.1K | Open weightsYes | LicenseApache 2.0 |
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Gemma 4 31B and its available data.
Gemma 4 31B is a large language model from Google based on the same research as Gemini 3. It uses Per-Layer Embeddings, Shared KV Cache, alternating sliding-window and global attention, and variable aspect ratio vision encoding.
Data as of 2026-09-08.
Common questions about Gemma 4 31B.
Gemma 4 31B's default version was released on Apr 2, 2026.
No official standard PAYG price is currently available for Gemma 4 31B. The lowest tracked third-party offer starts at $0.09 input and $0.34 output via OpenRouter.
Gemma 4 31B was created by Google.
The default version has a 262.1K token context window.
Yes. The default version is marked as open weight under Apache 2.0.
34 provider offerings are linked to the default version.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.