BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

Claude Opus 5 vs Gemini 3.1 Pro

Full profiles: Claude Opus 5 (Anthropic) and Gemini 3.1 Pro (Google): every benchmark, price and rank.

Updated · Data from 14 sources

As of 25 Oct 2026, Claude Opus 5 (max) leads on quality (68.8 vs 60.5 on the BenchLeader Index), but Gemini 3.1 Pro costs 2.2× less, so it is the better buy unless you need the extra quality. Pick Claude Opus 5 for coding (70 vs 57); pick Gemini 3.1 Pro for long context (66 vs 65).

Claude Opus 5 benchmarks max Anthropic Gemini 3.1 Pro benchmarks GoogleOpen in the comparison tool

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingClaude Opus 570 vs 57 category score; Arena WebDev 1691 vs 1447High
Agents & tool useClaude Opus 564 vs 57 category score; τ³-Banking (AA) 42.1% vs 21.4%Medium
MathsClaude Opus 571 vs 61 category score; FrontierMath Tiers 1–3 85.6% vs 59.6%High
ReasoningClaude Opus 574 vs 68 category score; ARC-AGI-2 90.4% vs 77.1%Medium
KnowledgeClaude Opus 572 vs 64 category score; AA-Omniscience 37.1 vs 31.9High
Long documentsGemini 3.1 Pro66 vs 65 category score; AA-LCR 82.0% vs 79.3%Medium
Images and other mediaClaude Opus 568 vs 65 category score; MMMU-Pro 84.7% vs 82.4%Medium
Chat quality (human preference)Claude Opus 569 vs 59 on Arena votesHigh
Lowest costGemini 3.1 Pro$4.50 vs $10.00 per 1M blended; agentic coding session $0.168 vs $0.400–
Fastest first answerEither24.1 s (Claude Opus 5 (max)) vs 21.1 s (Gemini 3.1 Pro) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • Claude Opus 5 (max) leads on quality: 68.8 vs 60.5.
  • Claude Opus 5 (max) is stronger in agents & tools, coding, composite, human preference, knowledge, maths, multimodal, reasoning.
  • Gemini 3.1 Pro is stronger in long context, instruction following.
  • Gemini 3.1 Pro is 2.2× cheaper ($4.50 vs $10.00 per 1M blended).
  • Gemini 3.1 Pro streams 2.3× faster (132 vs 57 tokens per second).

At a glance

SpecClaude Opus 5Gemini 3.1 Pro
DeveloperAnthropicGoogle
Released24 Jul 202619 Feb 2026
Context window (tokens)1,000,0001,048,576
Compared atClaude Opus 5 (max)Gemini 3.1 Pro
Open weightsNoNo
Price in / out / cached input per 1M$5.00 / $25.00 / $0.500$2.00 / $12.00 / $0.200

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryClaude Opus 5 (max)Gemini 3.1 ProLeader
Reasoning74.168.2Claude Opus 5
Coding70.256.9Claude Opus 5
Agents & tools64.357.1Claude Opus 5
Maths70.660.8Claude Opus 5
Knowledge71.663.9Claude Opus 5
Instruction following–67.5not comparable
Human preference68.558.7Claude Opus 5
Multimodal67.964.8Claude Opus 5
Long context64.766.1Gemini 3.1 Pro
Composite86.349.2Claude Opus 5

Where sources disagree

Across the 42 benchmarks both have, Claude Opus 5 leads 32, Gemini 3.1 Pro leads 9 and 1 is a tie. Every overall index that rates both puts Claude Opus 5 ahead.

Overall indexClaude Opus 5 (max)Gemini 3.1 ProLeader
BenchLeader Index68.860.5Claude Opus 5
AA Intelligence Index v4.3.250.829.7Claude Opus 5
Arena Text14911487Claude Opus 5

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.

MeasureClaude Opus 5 (max)Gemini 3.1 Pro
Blended price per 1M tokens$10.00$4.50
Input / output per 1M$5.00 / $25.00$2.00 / $12.00
Cached input per 1M$0.500$0.200
Price per task: chat reply$0.0095$0.0044
Price per task: summarize a 30-page report$0.075$0.031
Price per task: code edit$0.068$0.030
Price per task: agentic coding session$0.400$0.168
Price per task: structured extraction$0.015$0.0064
Cost to run a full benchmark suite$5.86$1.30
Output speed57 tok/s132 tok/s
Time to first answer24.1 s21.1 s
Time to a complete answer32.9 s24.9 s

Per token is not per task: on Artificial Analysis' full suite Claude Opus 5 (max) cost $5.86 and Gemini 3.1 Pro $1.30, because Claude Opus 5 (max) used about 2.0× as many tokens for the same work, which a per-token price hides.

Each model is compared at its best-ranked reasoning-effort setting. Claude Opus 5 (max) scores 68.8 and takes 24 s to its first answer token; Claude Opus 5 (high) scores 67.9 in 10 s, Claude Opus 5 (xhigh) scores 67.4 in 25 s. Gemini 3.1 Pro scores 60.5 and takes 21 s to its first answer token; Gemini 3.1 Pro (high) scores 60.3 in 21 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

168 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherClaude Opus 5 (max)Gemini 3.1 Pro
BenchLeader IndexBenchLeader68.860.5
Agents & tools scoreBenchLeader64.357.1
Coding scoreBenchLeader70.256.9
Composite scoreBenchLeader86.349.2
Human preference scoreBenchLeader68.558.7
Knowledge scoreBenchLeader71.663.9
Long context scoreBenchLeader64.766.1
Maths scoreBenchLeader70.660.8
Multimodal scoreBenchLeader67.964.8
Reasoning scoreBenchLeader74.168.2
Instruction following scoreBenchLeader–67.5
Blended price $/MBenchLeader$10.00$4.50
Output speedBenchLeader57 tok/s132 tok/s
Time to first answerBenchLeader24.1 s21.1 s
Context windowBenchLeader1,000,000 tokens1,048,576 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.9%94.1%
Humanity's Last ExamScale AI / CAIS–46.4%
SimpleBenchSimpleBench80.6%†79.6%
Arena Hard PromptsArena (LMArena)15151508
LiveBench 2.0 ReasoningLiveBench91.2%†84.0%
LiveBench 1.0 ReasoningLiveBench91.2%84.0%†
GPQA Diamond (AA)Artificial Analysis93.2%94.1%
Humanity's Last Exam (AA)Artificial Analysis54.9%47.0%
GPQA Diamond (Vals)Vals AI93.4%†95.5%†
EnigmaEvalScale AI SEAL–36.8%†
ARC-AGI-1ARC Prize97.5%98.0%
ARC-AGI-2ARC Prize90.4%77.1%
ARC-AGI-3ARC Prize30.2%†0.4%
CritPtArtificial Analysis29.1%17.7%
MysteryMechanismVals AI37.4%†20.7%†
Chess PuzzlesEpoch AI Benchmarking Hub43.0%51.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub59.0%34.0%†
CL-benchEpoch AI Benchmarking Hub–20.8%
CL-bench LifeEpoch AI Benchmarking Hub–16.9%
LMCAEpoch AI Benchmarking Hub63.3%53.8%†
DTBenchEpoch AI Benchmarking Hub97.6%95.5%†
ForecastBenchForecastBench66.5%64.3%
BTF-3Epoch AI Benchmarking Hub12.0%†–
Coding
SciCodeSciCode56.4%58.7%
WeirdMLWeirdML91.8%72.1%
Show all 168 benchmarks
MetricPublisherClaude Opus 5 (max)Gemini 3.1 Pro
Coding
FrontierCodeCognition53.4%–
GSO-BenchGSO-Bench–22.6%
Arena CodingArena (LMArena)15281522
Arena WebDevArena (LMArena)16911447
LiveBench 2.0 CodingLiveBench90.8%†66.7%
LiveBench 1.0 CodingLiveBench81.5%76.5%†
LiveCodeBenchVals AI89.0%†88.5%†
IOIVals AI84.3%†51.8%†
SWE-bench (Vals)Vals AI97.0%†78.8%†
SWE-Bench ProScale AI SEAL–46.1%†
Code MigrationVals AI57.5%†17.3%†
ProgramBenchVals AI3.0%†0.0%†
Vibe Code Bench 1-100Vals AI28.5%†6.7%†
Vibe Code Bench v1.1Vals AI88.4%†32.0%†
SWE-Bench Pro (private)Scale AI SEAL–46.1%†
SWE Atlas: Codebase QnAScale AI SEAL63.2%†13.5%
SWE Atlas: RefactoringScale AI SEAL–33.8%
SWE Atlas: Test WritingScale AI SEAL62.2%†29.8%
MirrorCodeEpoch AI Benchmarking Hub–8.9%†
DeepSWE v1.1Epoch AI Benchmarking Hub73.7%11.7%
CursorBenchCursor46.6%–
ALE-BenchEpoch AI Benchmarking Hub2164.6†1160.6
AlgoTuneAlgoTune–2.0
FrontierSWEEpoch AI Benchmarking Hub52.0%–
Terminal-Bench 4.0 (AA)Artificial Analysis49.0%4.0%
Terminal-Bench 2.1 (AA)Artificial Analysis89.1%73.8%
Startrise FrontendStartrise82.3%†–
Agents & tools
Terminal-BenchTerminal-Bench51.8%80.2%
OSWorld-Verified 2.0OSWorld31.4%–
APEX-AgentsMercor65.8%35.1%
Arena AgentArena (LMArena)8.2-4.4
LiveBench Long AgenticLiveBench62.1%†23.9%
LiveBench BrowserLiveBench84.0%†4.0%
LiveBench Agentic CodingLiveBench65.2%44.1%†
Terminal-Bench HardArtificial Analysis–53.8%
τ²-Bench Telecom (AA)Artificial Analysis–95.6%
Terminal-Bench 2.1 (Vals)Vals AI84.6%†70.8%†
MCP AtlasScale AI SEAL85.8%†78.2%†
HiL-BenchScale AI SEAL57.0%†35.3%
τ³-bench Bankingτ²-bench48.7%26.0%†
GDPval-AA v2.1Artificial Analysis–14.6%
τ³-Banking (AA)Artificial Analysis42.1%21.4%
ITBench SRE (AA)Artificial Analysis–30.3%
Analyst Agent (AA)Artificial Analysis53.8%41.3%
APEX-Agents (AA)Artificial Analysis–32.0%
CUA-benchVals AI9.0%–
CyberBenchVals AI65.4%†–
Finance Agent v2Vals AI58.6%†43.0%†
Harvey's Legal Agent Benchmark v1Vals AI6.7%†0.0%†
Legal Research BenchVals AI55.3%†20.7%†
SkillsBenchVals AI60.4%†–
SREBenchVals AI12.2%†0.0%†
Tax Agent BenchVals AI45.5%†9.3%†
Terminal-Bench 2.0 (Vals)Vals AI–67.4%†
Terminal-Bench 4.0 (Vals)Vals AI53.5%†2.5%†
Terminal-Bench ScienceVals AI27.1%†1.4%†
Time Horizon Index: KSPVals AI18.8%†–
Arena SearchArena (LMArena)1214†–
BALROGBALROG63.4%57.0%
DeepResearch BenchDeepResearch Bench–47.8%†
PostTrainBenchEpoch AI Benchmarking Hub35.0%†22.0%
ExploitBenchEpoch AI Benchmarking Hub–26.1%
METR Time HorizonsMETR–77.0%
GBAEvalEpoch AI Benchmarking Hub79.6%†0.8%
Vending-Bench 2Andon Labs11181.9†911.2
AutomationBenchArtificial Analysis56.6%35.4%
GDP.pdf (AA)Artificial Analysis21.6%17.8%
Harvey LAB-AA v1.1Harvey–0.0%
EnterpriseOps-GymArtificial Analysis47.5%42.2%
AA-Briefcase v1.1Artificial Analysis1657456
Terminal-Bench-Science 0.1 (AA)Artificial Analysis28.6%–
DeepSec-Bench (AA)Artificial Analysis25.9%–
BrowseCompOpenAI–31.2%
DeepSearchQAGoogle–60.2%
FACTS SearchGoogle85.3%†83.7%
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub85.6%59.6%
FrontierMath Tier 4Epoch AI Benchmarking Hub73.2%26.8%
OTIS Mock AIMEEpoch AI Benchmarking Hub98.9%95.6%
ProofBenchVals AI99.0%26.0%
LiveBench MathematicsLiveBench95.7%91.0%†
AIME (Vals)Vals AI–98.1%†
AIME 2026MathArena–98.3%
HMMT February 2026MathArena–94.7%
USAMO 2026MathArena–74.4%
MathArena ApexMathArena–60.9%
Arena MathsArena (LMArena)15171489
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub59.9%73.5%†
LiveBench 2.0 Data AnalysisLiveBench74.6%†78.5%
LiveBench ResearchLiveBench88.9%†65.7%
LiveBench 1.0 Data AnalysisLiveBench74.5%78.5%†
AA-OmniscienceArtificial Analysis37.131.9
MMLU-ProVals AI91.6%†91.0%†
LegalBenchVals AI87.0%†87.4%†
CorpFinVals AI73.2%†64.5%†
TaxEvalVals AI75.1%†72.9%†
MedQAVals AI–96.4%†
PRBench FinanceScale AI SEAL–41.9%
PRBench LegalScale AI SEAL–44.0%
MultiNRCScale AI SEAL–64.7%
BioMysteryBenchVals AI79.3%†60.0%†
CaseLaw v2Vals AI–64.8%†
Excel Modeling BenchmarkVals AI73.6%†52.6%†
MedCodeVals AI63.6%†59.1%†
MedScribeVals AI91.0%†76.1%†
Public Benefits Bench v1.1Vals AI76.9%†53.8%†
DrugDiscoveryBenchScale AI SEAL53.7%41.9%†
EBR-benchEpoch AI Benchmarking Hub45.7%14.3%
GDP.pdfEpoch AI Benchmarking Hub24.0%17.0%
AA-Omniscience: accuracyArtificial Analysis60.9%54.9%
AA-Omniscience: non-hallucinationArtificial Analysis39.2%49.1%
Global-MMLU-Lite (AA)Artificial Analysis–93.2%
Instruction following
LiveBench LanguageLiveBench88.7%85.4%†
LiveBench Instruction FollowingLiveBench63.8%79.1%†
IFBenchArtificial Analysis–77.1%
MultiChallengeScale AI SEAL–71.4%
TutorBenchScale AI SEAL–53.0%
Arena Instruction FollowingArena (LMArena)14951480
Human preference
Arena TextArena (LMArena)14911487
EQ-Bench 4EQ-Bench1360†1136
Arena Creative WritingArena (LMArena)14741481
Arena Multi-turnArena (LMArena)14871496
Arena Longer QueriesArena (LMArena)15031499
Multimodal
Arena VisionArena (LMArena)1320†1296
LiveBench DesignLiveBench83.0%†61.6%
MMMU-ProArtificial Analysis84.7%82.4%
MMMU-Pro (official)MMMU–80.5%†
MMMU-Pro (Vals)Vals AI89.9%†88.2%†
MortgageTaxVals AI72.1%†69.4%†
SAGEVals AI49.4%†48.7%†
VTBScale AI SEAL–29.0%
Blueprint-Bench 2Epoch AI Benchmarking Hub30.4%†26.5%
Long context
AA-LCRArtificial Analysis79.3%82.0%
Arena DocumentArena (LMArena)1490†1459
MLCRArtificial Analysis11.1%11.1%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub163.9†155.3
LiveBench 2.0LiveBench82.1%†54.9%
LiveBench 1.0LiveBench80.1%77.0%†
AA Intelligence Index v4.3.2Artificial Analysis50.829.7
Vals RSI IndexVals AI33.0%–
Vals Science IndexVals AI47.9%†27.4%†
Vals Multimodal IndexVals AI73.9%†56.1%†
Safety & honesty
FORTRESSScale AI SEAL16.8%†29.8%
MASKScale AI SEAL–42.4%

Data as of 2026-10-25. Best configuration of each model; every score links to its source on the model pages.

What changed

  • Gemini 3.1 Pro: 2 score revisions.
  • Claude Opus 5 (max): 1 new result (Vals RSI Index).
  • Claude Opus 5 (max): 1 score revision.
  • Gemini 3.1 Pro: 2 score revisions.
  • Claude Opus 5 (max): 1 new result (ForecastBench); 1 score revision.
  • Gemini 3.1 Pro: 1 score revision.
  • Claude Opus 5 (max): BenchLeader Index 68.3 → 68.9.
  • Gemini 3.1 Pro: BenchLeader Index 62.4 → 60.7.
  • Gemini 3.1 Pro: 8 new results (LiveBench 2.0, LiveBench Long Agentic, LiveBench 2.0 Reasoning…).
  • Claude Opus 5 (max): 4 new results (GDPval-AA v2.1, AA-Omniscience: accuracy, AA-Omniscience: non-hallucination…).

Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs

Claude Opus 5 vs Gemini 3.1 Pro: questions

Is Claude Opus 5 better than Gemini 3.1 Pro?
Claude Opus 5 (max) leads on quality: 68.8 vs 60.5. The BenchLeader Index combines every independent quality benchmark; Claude Opus 5 (max) is ahead overall as of 2026-10-25, but check the category scores for your use.
Which is cheaper, Claude Opus 5 or Gemini 3.1 Pro, and will that change?
Gemini 3.1 Pro is cheaper today: $4.50 against $10.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Claude Opus 5 better than Gemini 3.1 Pro for coding?
Claude Opus 5 scores higher in coding (70 vs 57 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1691 for Claude Opus 5 against 1447 for Gemini 3.1 Pro.
Is Claude Opus 5 better than Gemini 3.1 Pro for agents and tool use?
Claude Opus 5 scores higher in agentic tasks (64 vs 57 on the category index, where 50 is average). On Terminal-Bench, the top shared agents & tools benchmark, it is 51.8% for Claude Opus 5 against 80.2% for Gemini 3.1 Pro.
Which is better for long documents?
Gemini 3.1 Pro scores higher on long-context benchmarks (66 vs 65). Both take about 1M tokens: Claude Opus 5 1,000,000, Gemini 3.1 Pro 1,048,576.
Which has lower latency?
Gemini 3.1 Pro answers first: 21.1 s against 24.1 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Which is faster, Claude Opus 5 or Gemini 3.1 Pro?
Gemini 3.1 Pro streams faster: 132 against 57 output tokens per second.
Do independent benchmarks agree on Claude Opus 5 vs Gemini 3.1 Pro?
Across the 42 benchmarks both have, Claude Opus 5 leads 32, Gemini 3.1 Pro leads 9 and 1 is a tie. Every overall index that rates both puts Claude Opus 5 ahead.
When was Claude Opus 5 released?
Claude Opus 5 was released on 24 Jul 2026.