Gemma 4 31B

Gemma 4 31B is a reasoning model from Google in the Gemma 4 family. 24 benchmarks count toward its score, in 8 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.48.8 ±3.3
CoverageShare of the index weight with results.100%
SpeedOutput tokens per second.35/s
Input / 1MUS dollars per 1M input tokens.Free
Output / 1MUS dollars per 1M output tokens.Free
ContextMaximum tokens in one request.256K
EloLMArena rating and rank.1442 (#64)

The index is a score out of 100. The ± range shows how much it can change. A free tier is available.

5,894 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
43.7
CodingCode writing and repair.
52.8
ReasoningLogic problems and puzzles.
42.6
MultimodalTasks with images and text.
54.6
KnowledgeFacts and expert knowledge.
47.2
MultilingualTasks in many languages.
71.6
InstructionTasks with strict rules in the prompt.
65.8
MathMath problems.
52.6

Results

24 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
Artificial Analysis GPQA DiamondKnowledge85.7%56.4Artificial Analysis
Massive Multitask Language Understanding ProfessionalKnowledge85.2%54.8Yubo Wang et al.
Graduate-Level Google-Proof Q&AKnowledge84.3%56.1David Rein et al.
Massive Multi-discipline Multimodal Understanding ProMultimodal76.9%52.5MMMU-Pro authors
GPQA diamondKnowledge75.8%48.2minimal effortEpoch AI
Artificial Analysis IFBenchInstruction75.6%65.8Artificial Analysis
React Native EvalsCoding75.2%52.9Callstack
Artificial Analysis MMMU-ProMultimodal73.4%56.8Artificial Analysis
OTIS Mock AIME 2024-2025Math73.3%52.6minimal effortEpoch AI
Artificial Analysis Long Context ReasoningReasoning69.7%56.5Artificial Analysis
EuroEval FrenchMultilingual67.9%87.0EuroEval
EuroEval PortugueseMultilingual67.5%86.5EuroEval
EuroEval ItalianMultilingual66.8%85.7EuroEval
EuroEval SwedishMultilingual63.1%81.1EuroEval
EuroEval DutchMultilingual62.7%80.5EuroEval
EuroEval PolishMultilingual61.6%79.2EuroEval
τ²-Bench Tool-Agent-User EvaluationAgentic59.9%41.8Victor Barres et al.
EuroEval SpanishMultilingual55.8%71.9EuroEval
EuroEval GermanMultilingual55.7%71.8EuroEval
EuroEval PortugueseMultilingual55.3%71.3EuroEval
EuroEval DutchMultilingual51.8%67.0EuroEval
EuroEval ItalianMultilingual50.3%65.1EuroEval
EuroEval SwedishMultilingual50.3%65.1EuroEval
EuroEval FrenchMultilingual50.0%64.8EuroEval
EuroEval PolishMultilingual46.7%60.6EuroEval
Artificial Analysis SciCodeCoding45.5%56.0Artificial Analysis
Artificial Analysis Coding IndexCoding43.4%49.6Artificial Analysis
EuroEval GermanMultilingual42.6%55.6EuroEval
SWE-RebenchCoding41.6%Nebius
EuroEval SpanishMultilingual40.8%53.2EuroEval
Terminal-Bench 2.0Agentic39.3%50.74 Jun 2026Vals AI
Gert Labs Composite Game BenchmarkAgentic35.3%43.9Gert Labs
Humanity's Last ExamKnowledge26.5%51.2Center for AI Safety et al.
Artificial Analysis Humanity's Last ExamKnowledge23.6%50.1Artificial Analysis
Artificial Analysis Omniscience AccuracyKnowledge20.0%38.6Artificial Analysis
Humanity's Last Exam without toolsKnowledge19.5%45.3OpenAI
Artificial Analysis Intelligence IndexKnowledge19.0%46.1Artificial Analysis
SimpleQA VerifiedKnowledge10.4%31.2Epoch AI
Artificial Analysis Agentic IndexAgentic6.7%42.6Artificial Analysis
GDPval-AA normalizedAgentic5.3%39.6Artificial Analysis
Chess PuzzlesReasoning5.0%30.0minimal effortEpoch AI
Critical Physics TasksReasoning1.4%41.3Artificial Analysis

24 benchmarks count, from 41 of 42 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsEpoch AI, collected directlyCC BY — free to use and redistribute with attributionEuroEval, collected directlyMIT — the leaderboard site and its CSV routes are in the licensed repositoryVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AI

More from Google

Gemini 3.7 Flash72.3Gemini 3.8 Flash72.1Gemini 3.5 Flash67.3Gemini 3.6 Flash66.8Gemini 3.1 Pro64.7