Data as of November 13, 2026 · How the score is built
Decision readingGemma 4 31B scores 28.8 out of 100 and ranks #175 of 215. This profile shows 16 source-displayable benchmark rows; its strongest eligible category is Instruction Following at #16. Published weights can be self-hosted, but infrastructure cost varies and is not a comparable API token rate.
Email me when its price, rank or availability changes.
Capability
28.8/100
field median 49.4#175 of 215 ranked models
Public
#175of 215
Verified —
Price
Self-hosted; infrastructure cost varies
input median $0.60No comparable first-party hosted token rate
Speed
35tok/s
field median 92 tok/sFirst token 50.03 s
Context
256Ktokens
field median 256,000Maximum output length is tracked separately
Decision snapshot
Where this model sits in the ranked field, its strongest published evidence, and what to validate before choosing it. The marker compares it with the current ranked catalog, not with an absolute quality threshold.
Strongest published evidence
Instruction Following ranks #16. A well-rounded choice across a range of tasks.
Validate before choosing
16 published rows leave some tracked benchmark slots empty. No comparable first-party API token rate is published.
Source-linked · 16 displayable benchmark rows
Is Gemma 4 31B right for your job?
Its place on the LLM Selector’s job rankings, the same order the job pages publish. Describe your own job to see where it stands for that.
Where Gemma 4 31B ranks by benchmark score:
We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy
Category score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #133 of 135Percentile 1stWeight 22%3 benchmarksMixed sources | 12.5 | 3 benchmarks | Mixed sources | |||
| CodingRank #107 of 162Percentile 34thWeight 20%3 benchmarksVerified | 28.8 | 3 benchmarks | Verified | |||
| ReasoningRank #109 of 160Percentile 32ndWeight 17%1 benchmarkVerified | 32.0 | 1 benchmark | Verified | |||
| MultimodalRank #60 of 67Percentile 11thWeight 12%2 benchmarksVerified | 47.3 | 2 benchmarks | Verified | |||
| KnowledgeRank #121 of 180Percentile 33rdWeight 12%4 benchmarksMixed sources | 36.7 | 4 benchmarks | Mixed sources | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingRank #16 of 157Percentile 90thWeight 5%Not listedUnlisted sources | 92.9 | Not listed | Unlisted sources | |||
| MathWeight 5%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured |
16 of 919 tracked benchmark slots have displayable evidence · bars run 0–100
Coverage detailsHow much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic1/3 verified
- Coding3/3 verified
- Reasoning1/1 verified
- Multimodal2/2 verified
- Knowledge3/4 verified
- MultilingualNot measured
- Inst. FollowingNot measured
- MathNot measured
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
Gemma 4 31B category percentile values
- Agentic1st percentile
- Coding34th percentile
- Reasoning32nd percentile
- Multimodal11th percentile
- Knowledge33rd percentile
- MultilingualNot eligible
- Instruction following90th percentile
- MathNot eligible
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#133/135
- Coding#107/162
- Reasoning#109/160
- Multimodal#60/67
- Knowledge#121/180
- MultilingualNot ranked
- Inst. Following#16/157
- MathNot ranked
Benchmark ledger
Every result says who ran the test. For this model, Google ran 6 of 16. Coding opens by default. The tick marks the best source-verified result in the catalog; provisional leaders do not set the reference.
Coding3 rows
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| SWE-Rebench | 41.6% | Best verified: Claude Opus 4.6 · 65.3% | Gap23.7 behind | Run by SWE-Rebench | Counts5% ref. weight |
| LiveCodeBench v6 | 80.0% | Best verified: Sakana Fugu-Ultra · 93.2% | Gap13.2 behind | Reported by Google | Shown only |
| React Native Evals | 75.2% | Best verified: Composer 2 · 96.1% | Gap20.9 behind | Run by React Native Evals | Shown only |
Agentic3 rows
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| b3 Agent Securityb3: Backbone Breaker Benchmark | 83.0% | Best verified: Mistral Large 4 · 97.1% | Gap14.1 behind | Reported by a third party | Shown only |
| Cyber Misuse RefusalRefusal of harmful cyber requests (JailbreakBench, StrongREJECT, AgentHarm) | 84.0% | Best verified: Mistral Large 4 · 96.6% | Gap12.6 behind | Reported by a third party | Shown only |
| Gert LabsGert Labs Composite Game Benchmark | 35.26% | Best verified: Claude Opus 4.8 · 72.97% | Gap37.7 behind | Run by Gert Labs | Shown only |
Reasoning1 row
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| MRCRv2 | 66.4% | Best verified: Sakana Fugu-Ultra · 93.6% | Gap27.2 behind | Reported by Google | Counts20% ref. weight |
Multimodal2 rows
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| MMMU-ProMassive Multi-discipline Multimodal Understanding Pro | 76.9% | Best verified: GPT-6.1 Terra · 85.5% | Gap8.6 behind | Reported by Google | Counts30% ref. weight |
| ParseBench-1ParseBench-1: document parsing for AI agents (LlamaIndex leaderboard, first release) | 62.40% | Best verified: Claude Opus 5.5 · 78.80% | Gap16.4 behind | Run by ParseBench-1 | Counts7% ref. weight |
Knowledge4 rows
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| HLEHumanity's Last Exam | 26.5% | Best verified: Claude Fable 5.2 · 74.2% | Gap47.7 behind | Reported by Google | Counts44% ref. weight |
| HLE w/o toolsHumanity's Last Exam without tools | 19.5% | Best verified: Claude Fable 5.2 · 73.2% | Gap53.7 behind | Reported by Google | Counts7% ref. weight |
| MMLU-ProMassive Multitask Language Understanding Professional | 85.2% | Best verified: Qwen3.7 Max · 89.6% | Gap4.4 behind | Reported by Google | Counts6% ref. weight |
| GPQAGraduate-Level Google-Proof Q&A | 84.3% | Best verified: GPT-6 Astra · 96% | Gap11.7 behind | Carried from an upstream page | Counts3% ref. weight |
External signals3 rows
| Benchmark | Score | Versus best verified row | Gap | Who ran it | Use in ranking |
|---|---|---|---|---|---|
| NYT ConnectionsNYT Connections LLM Benchmark (extended version) | 70.6% | Best verified: Gemini 4 Carbon · 99.3% | Gap28.7 behind | Run by NYT Connections | Shown only |
| FACTS Multimodal | 42.3% | Best verified: Gemini 4 Carbon · 60.2% | Gap17.9 behind | Run by FACTS Multimodal | Shown only |
| StealthMarkStealthMark Intelligence Benchmarks (overall score) | 29.5 | Best verified: GPT-6 Astra · 74.6 | Gap45.1 behind | Run by StealthMark | Shown only |
Outlined bars are lab-reported or carried from an upstream page. Filled bars are benchmark-run or independent.
Lineage
The sequence follows explicit supersedes links. Each score is estimated for that model; a relative can inform a sparse estimate but never sets a floor, so a newer release can score below an earlier one. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- gemma-4-31b-itGoogle Gemma Gemini API guide
- Context window
- 256KGoogle Gemma 4 model documentation
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- text, imageGoogle Gemma 4 model documentation
- Output modalities
- textGoogle Gemma 4 model documentation
- Parameters
- Not sourced yet
- Availability
- Gemini API · Google AI Studio · open weightsGoogle Gemma Gemini API guide
- Cloud regions
- Not tracked yet
- Lifecycle
- activeGoogle gemma-4-31B model card
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing record
- Self-host
- Gemma 4 31B needs ~24GB VRAM (1× NVIDIA RTX 4090 (24GB)).
- Rate limits
- Not tracked yet
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
Gemma 4 31B ranks #175 of 215 on the public leaderboard with a score of 28.81/100. It does not yet have enough sourced coverage for a verified position.
Gemma 4 31B is a open weight model with a 256K context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Official exact-value snapshot from Google's April 2, 2026 Gemma 4 model card and launch post. BenchLM maps only schema-equivalent results from the published table: MMLU Pro, GPQA Diamond, LiveCodeBench v6, HLE no tools / with search, MMMU Pro, and MRCR v2. AIME 2026, Codeforces, MMMLU, OmniDocBench 1.5, MATH-Vision, MedXPertQA MM, audio benchmarks, BigBench Extra Hard (no matching key), and Tau2 (average over 3) remain out-of-schema or non-equivalent to current weighted keys.
Gemma 4 31B sits in the Gemma 4 family with Gemma 4 12B, Gemma 4 26B A4B, Gemma 4 E4B, Gemma 4 E2B. 16 of 919 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Instruction Following at #16, while its lowest eligible position is Agentic at #133. a well-rounded choice across a range of tasks.
Last updated November 13, 2026. Runtime fields remain blank until a sourced snapshot exists.
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Where to run Gemma 4 31B by API
4 hosts publish a per-token price for it. DeepInfra has the lowest output rate, $0.34 per million tokens. Prices read from each host’s own pages on November 11, 2026.
| Host | Input /1M | Output /1M | Cached /1M | Context |
|---|---|---|---|---|
| DeepInfraServes gemma-4-31B-it-turbo | $0.09 | $0.34 | $0.05 | 262K |
| Novita AI | $0.14 | $0.40 | — | 262K |
| OpenRouter | $0.20 | $0.40 | — | 262K |
| SambaNovaServes Gemma 4 31B IT | $0.38 | $1.15 | — | 128K |
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Questions
How does Gemma 4 31B perform overall in AI benchmarks?
Gemma 4 31B ranks #175 out of 215 models on the public BenchAlign leaderboard, with a score of 28.81/100. Its evidence status is Estimated, and this profile shows 16 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is Gemma 4 31B good for knowledge and understanding?
Gemma 4 31B ranks #121 out of 180 eligible models for knowledge and understanding, with a public category score of 36.7/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Gemma 4 31B good for coding and programming?
Gemma 4 31B ranks #107 out of 162 eligible models for coding and programming, with a public category score of 28.8/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Gemma 4 31B good for reasoning and logic?
Gemma 4 31B ranks #109 out of 160 eligible models for reasoning and logic, with a public category score of 32/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Gemma 4 31B good for agentic tool use and computer tasks?
Gemma 4 31B ranks #133 out of 135 eligible models for agentic tool use and computer tasks, with a public category score of 12.5/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Gemma 4 31B good for multimodal and grounded tasks?
Gemma 4 31B ranks #60 out of 67 eligible models for multimodal and grounded tasks, with a public category score of 47.3/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Gemma 4 31B open source?
Gemma 4 31B is an open-weight model from Google. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Which sibling models are related to Gemma 4 31B?
Gemma 4 31B belongs to the Gemma 4 family. Related tracked variants include Gemma 4 12B, Gemma 4 26B A4B, Gemma 4 E4B, plus 1 more. A sibling link indicates shared lineage or a documented configuration relationship; it does not mean the variants have identical pricing, context limits, benchmark evidence, or deployment behavior. Compare before switching.
Does Gemma 4 31B have full benchmark coverage on BenchLM?
No. Gemma 4 31B currently has 33 source-displayable rows. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of Gemma 4 31B?
Gemma 4 31B has a documented context window of 256K. That figure is the maximum combined prompt and retained-conversation space reported for this exact model; it is not the maximum output length. The profile keeps output limits separate because providers often publish those limits independently.