Gemini 3.1 Pro Preview

Gemini 3.1 Pro Preview is a model from Google. 16 benchmarks count toward its score, in 6 categories.

availableShows if the model has enough results for an index.
IndexOverall score. 50 is the middle.61.5 ±5.0
CoverageShare of the index weight with results.90%
SpeedOutput tokens per second.90/s
Input / 1MUS dollars per 1M input tokens.$2 batch $1
Output / 1MUS dollars per 1M output tokens.$12 batch $6 US dollars per 1M output tokens in a batch.
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1480 (#16)

50 is the middle of the board. The range shows the doubt in the index. Batch work costs less.

106,951 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category. 50 is the middle.

50 is the middle
AgenticMulti-step tasks with tools.
53.8
CodingCode writing and repair.
59.2
ReasoningLogic problems and puzzles.
76.0
MultimodalTasks with images and text.
70.9
KnowledgeFacts and expert knowledge.
65.1
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
N/A
MathMath problems.
59.7

Results

16 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result on the index scale.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
AIMEMath98.1%59.516 Apr 2026Vals AI
ARC-AGI-1 (semi-private)Reasoning98.0%73.5ARC Prize Foundation
GPQA DiamondKnowledge95.5%66.41 Sept 2026Vals AI
MMLU ProKnowledge91.0%63.91 Sept 2026Vals AI
LiveCodeBenchCoding88.5%64.81 Sept 2026Vals AI
MMMU ProMultimodal88.2%70.91 Sept 2026Vals AI
SWE-benchCoding78.8%60.71 Sept 2026Vals AI
ARC-AGI-2 (semi-private)Reasoning77.1%78.6ARC Prize Foundation
Terminal-Bench 2.1Agentic70.8%65.821 Sept 2026Vals AI
Terminal-Bench 2.0Agentic67.4%70.84 Jun 2026Vals AI
Vibe Code Bench v1.1Coding32.0%55.5OpenHands21 Sept 2026Vals AI
τ²-bench BankingAgentic26.0%17.5high effort · Sierra4 Aug 2026Sierra Research
ProofBench v1.1Math26.0%59.921 Sept 2026Vals AI
Code MigrationCoding17.3%56.021 Sept 2026Vals AI
Vibe Code Bench 1-100Coding6.7%59.1OpenHands16 Sept 2026Vals AI
ARC-AGI-3 (semi-private)Reasoning0.4%ARC Prize Foundation
ProgramBenchCoding0.0%21 Sept 2026Vals AI
Agent Arena steerabilityAgentic-3.663.415 Sept 2026LMArena
Agent Arena task outcomeAgentic-5.561.215 Sept 2026LMArena
Agent Arena command recoveryAgentic-20.444.415 Sept 2026LMArena

16 benchmarks count, from 18 of 20 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 authorsOpenRouter, collected directlyNo licence statedVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AIARC Prize Foundation, collected directlyNo licence stated. Their terms ask for written permission before commercial useSierra Research, collected directlyMIT — results are in the licensed repositoryLMArena, collected directlyCC BY 4.0 (lmarena-ai/leaderboard-dataset on Hugging Face)

Same level, lower price

Qwen3.8-Flash-Next66.4 · Freedots3-note Preview65.5 · FreeApodex 1.1 Mini59.1 · Free

More from Google

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