Gemini 3 Pro

Gemini 3 Pro is a non-reasoning model from Google. 32 benchmarks count toward its score, in 8 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.60.5 ±2.9
CoverageShare of the index weight with results.100%
SpeedOutput tokens per second.109/s
Input / 1MUS dollars per 1M input tokens.$2
Output / 1MUS dollars per 1M output tokens.$12
ContextMaximum tokens in one request.2M
EloLMArena rating and rank.1479 (#17)

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

40,654 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.
56.8
CodingCode writing and repair.
57.4
ReasoningLogic problems and puzzles.
60.0
MultimodalTasks with images and text.
57.9
KnowledgeFacts and expert knowledge.
65.9
MultilingualTasks in many languages.
83.2
InstructionTasks with strict rules in the prompt.
60.5
MathMath problems.
60.1

Results

32 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.
AIMEMath96.7%58.816 Apr 2026Vals AI
MATH 500Math96.4%49.59 Jan 2026Vals AI
MGSMMultilingual93.9%9 Jan 2026Vals AI
GPQA diamondKnowledge92.6%63.7Epoch AI
Artificial Analysis Global-MMLU-LiteMultilingual92.2%Artificial Analysis
Artificial Analysis LiveCodeBenchCoding91.7%Artificial Analysis
GPQA DiamondKnowledge91.7%62.91 Sept 2026Vals AI
OTIS Mock AIME 2024-2025Math91.4%62.6Epoch AI
τ²-bench TelecomAgentic91.0%64.1high effort · Sierra2 Mar 2026Sierra Research
Artificial Analysis GPQA DiamondKnowledge90.8%61.6Artificial Analysis
MMLU ProKnowledge90.1%62.51 Sept 2026Vals AI
Artificial Analysis MMLU-ProKnowledge89.8%Artificial Analysis
V*Multimodal88.0%46.7Z.AI
VideoMMMUMultimodal87.6%Qwen
MMMU ProMultimodal87.5%69.81 Sept 2026Vals AI
τ²-Bench Tool-Agent-User EvaluationAgentic87.1%61.3Victor Barres et al.
MathVisionMultimodal86.6%Qwen
LiveCodeBenchCoding86.4%62.91 Sept 2026Vals AI
CharXiv ReasoningMultimodal81.4%58.8CharXiv authors
Massive Multi-discipline Multimodal Understanding ProMultimodal81.0%59.2MMMU-Pro authors
τ²-bench AirlineAgentic80.5%56.6high effort · Sierra2 Mar 2026Sierra Research
Artificial Analysis MMMU-ProMultimodal80.2%65.1Artificial Analysis
SWE-benchCoding76.4%58.81 Sept 2026Vals AI
Artificial Analysis Long Context ReasoningReasoning76.0%60.8Artificial Analysis
τ²-bench RetailAgentic75.9%53.3high effort · Sierra30 Apr 2026Sierra Research
ARC-AGI-1 (semi-private)Reasoning75.0%62.6ARC Prize Foundation
SWE-bench VerifiedCoding74.2%57.1mini-SWE-agent26 Feb 2026SWE-bench team
SWE-Bench verifiedCoding72.9%56.0Epoch AI
ScreenSpot ProMultimodal72.7%54.3Kaixin Li et al.
Artificial Analysis IFBenchInstruction70.4%60.5Artificial Analysis
SWE-bench VerifiedCoding69.6%53.4high effort · mini-SWE-agent1 Sept 2026SWE-bench team
SWE-bench MultilingualMultilingual68.7%mini-SWE-agent20 Feb 2026SWE-bench team
SWE-bench MultilingualCoding68.7%mini-SWE-agent2 Sept 2026SWE-bench team
EuroEval PolishMultilingual64.8%83.2EuroEval
Gert Labs Composite Game BenchmarkAgentic63.2%68.6Gert Labs
Artificial Analysis Omniscience AccuracyKnowledge55.8%82.9Artificial Analysis
Terminal-Bench 2.0Agentic55.1%62.04 Jun 2026Vals AI
Terminal-Bench 1.0Agentic51.3%55.012 Jan 2026Vals AI
Artificial Analysis Humanity's Last ExamKnowledge39.7%67.6Artificial Analysis
IOI v1Coding38.8%62.09 Aug 2026Vals AI
FrontierMath-2025-02-28-PrivateMath37.6%65.7Epoch AI
ARC-AGI-2 (semi-private)Reasoning31.1%55.4ARC Prize Foundation
Chess PuzzlesReasoning31.0%63.6Epoch AI
Artificial Analysis Intelligence IndexKnowledge28.0%57.4Artificial Analysis
FrontierMath-Tier-4-2025-07-01-PrivateMath18.8%63.8Epoch AI
τ²-bench BankingAgentic18.0%11.8high effort · Sierra4 Aug 2026Sierra Research
Vibe Code Bench v1.1Coding14.3%48.1OpenHands21 Sept 2026Vals AI
JobBenchAgentic11.4%42.0Yuetai Li et al.
Critical Physics TasksReasoning9.1%57.5Artificial Analysis

32 benchmarks count, from 41 of 49 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 authorsVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AIEpoch AI, collected directlyCC BY — free to use and redistribute with attributionSierra Research, collected directlyMIT — results are in the licensed repositoryARC Prize Foundation, collected directlyNo licence stated. Their terms ask for written permission before commercial useSWE-bench team, collected directlyNo licence stated for the leaderboard. The harness repo is MITEuroEval, collected directlyMIT — the leaderboard site and its CSV routes are in the licensed repository

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