BenchLeader

Claude Opus 5 vs Gemini 3.8 Flash

Full profiles: Claude Opus 5 (Anthropic) and Gemini 3.8 Flash (Google): every benchmark, price and rank.

Updated · Data from 15 sources

As of 10 Nov 2026, Claude Opus 5 is ahead of Gemini 3.8 Flash on quality (65.2 vs 61.8 on the BenchLeader Index), but Gemini 3.8 Flash is 6.7× cheaper at $1.50 vs $10.00 per 1M tokens blended. Claude Opus 5 leads in every category we measure both in. Gemini 3.8 Flash's current price runs until 31 Dec 2026, then $3.00.

Key facts (as of 10 Nov 2026)

  • Claude Opus 5 scores 65.2 on the BenchLeader Index and Gemini 3.8 Flash scores 61.8 (BenchLeader, 10 Nov 2026).
  • Claude Opus 5 costs $10.00 per 1M tokens blended (list price, 10 Nov 2026).
  • Gemini 3.8 Flash costs $1.50 per 1M tokens blended until 31 Dec 2026, then $3.00 (https://ai.google.dev/gemini-api/docs/pricing, checked 20 Oct 2026).
  • Gemini 3.8 Flash streams ~1.5× faster than Claude Opus 5 (Artificial Analysis, 7-day median, 10 Nov 2026).
  • Of the 35 benchmarks both models have, Claude Opus 5 leads 26 (BenchLeader, 10 Nov 2026).

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 568 vs 60 category score; WeirdML 86.3% vs 84.8%High
Agents & tool useClaude Opus 565 vs 51 category score; Terminal-Bench 2.1 (Vals) 84.6% vs 81.3%High
MathsClaude Opus 566 vs 62 category score; OTIS Mock AIME 97.8% vs 98.9%Medium
ReasoningClaude Opus 571 vs 63 category score; GPQA Diamond 92.9% vs 95.4%High
KnowledgeClaude Opus 566 vs 64 category score; MMLU-Pro 91.6% vs 90.2%Medium
Chat quality (human preference)Claude Opus 576 vs 69 on Arena votesMedium
Lowest costGemini 3.8 Flash$1.50 vs $10.00 per 1M blended; agentic coding session $0.060 vs $0.400; from 1 Jan 2027: $3.00 vs $10.00–
Fastest first answerClaude Opus 52.8 s (Claude Opus 5) vs 16.6 s (Gemini 3.8 Flash (high)) 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 leads on quality: 65.2 vs 61.8.
  • Claude Opus 5 is stronger in agents & tools, coding, human preference, knowledge, maths, reasoning.
  • Gemini 3.8 Flash (high) is stronger in long context, multimodal.
  • Gemini 3.8 Flash (high) is 6.7× cheaper ($1.50 vs $10.00 per 1M blended) until 31 Dec 2026, then $3.00 vs $10.00 from 1 Jan 2027.
  • Gemini 3.8 Flash (high) streams 1.8× faster (112 vs 63 tokens per second).

At a glance

SpecClaude Opus 5Gemini 3.8 Flash
DeveloperAnthropicGoogle
Released24 Jul 20262 Sep 2026
Context window (tokens)1,000,0001,048,576
Compared atClaude Opus 5Gemini 3.8 Flash (high)
Open weightsNoNo
Price in / out / cached input per 1M$5.00 / $25.00 / $0.500$0.750 / $3.75 / $0.075
Announced price changeNone announced$1.50 / $7.50 from 1 Jan 2027

Limits, availability and announced prices from the vendors' own pages: ai.google.dev pricing (checked 20 Oct 2026). Everything else is from the day's data.

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 5Gemini 3.8 Flash (high)Leader
Reasoning71.263.2Claude Opus 5
Coding67.960.2Claude Opus 5
Agents & tools65.151.0Claude Opus 5
Maths65.761.8Claude Opus 5
Knowledge66.364.1Claude Opus 5
Human preference75.969.2Claude Opus 5
Multimodal–67.5not comparable
Long context–65.4not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsKnowledgeHumanpreferenceClaude Opus 5, Reasoning: 71.2Claude Opus 5, Coding: 67.9Claude Opus 5, Agents & tools: 65.1Claude Opus 5, Maths: 65.7Claude Opus 5, Knowledge: 66.3Claude Opus 5, Human preference: 75.9Gemini 3.8 Flash, Reasoning: 63.2Gemini 3.8 Flash, Coding: 60.2Gemini 3.8 Flash, Agents & tools: 51.0Gemini 3.8 Flash, Maths: 61.8Gemini 3.8 Flash, Knowledge: 64.1Gemini 3.8 Flash, Human preference: 69.2

Not drawn, measured for fewer than two of these models: Instruction following, Long context, Multimodal.

ModelIndex

Where sources disagree

Across the 35 benchmarks both have, Claude Opus 5 leads 26, Gemini 3.8 Flash leads 8 and 1 is a tie. Independent sources split: Claude Opus 5 leads on BenchLeader Index and AA Intelligence Index v5.0; Gemini 3.8 Flash leads on Arena Text.

Overall indexClaude Opus 5Gemini 3.8 Flash (high)Leader
BenchLeader Index65.261.8Claude Opus 5
AA Intelligence Index v5.044.2†30.9Claude Opus 5
Arena Text1490†1498Gemini 3.8 Flash

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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureClaude Opus 5Gemini 3.8 Flash (high)
Blended price per 1M tokens, until 31 Dec 2026$10.00$1.50
Input / output per 1M$5.00 / $25.00$0.750 / $3.75
Blended price per 1M from 1 Jan 2027$10.00$3.00
Input / output per 1M from 1 Jan 2027$5.00 / $25.00$1.50 / $7.50
Cached input per 1M$0.500$0.075
Price per task: chat reply$0.0095$0.0014
Price per task: summarize a 30-page report$0.075$0.011
Price per task: code edit$0.068$0.010
Price per task: agentic coding session$0.400$0.060
Price per task: structured extraction$0.015$0.0023
Agentic coding session, input mostly cached$0.198$0.030
Cost per benchmark task (AA Index)–$7.77
Output tokens per task (AA Index)–99k
Of which reasoning–60k
Time per task (AA Index)–15 min
Output speed63 tok/s112 tok/s
Time to first answer2.8 s16.6 s
Time to a complete answer–50.3 s

Each model is compared at its best-ranked reasoning-effort setting. Claude Opus 5 scores 65.2 and takes 3 s to its first answer token; Claude Opus 5 (max) scores 65.0 in 48 s, Claude Opus 5 (high) scores 64.2 in 10 s. Gemini 3.8 Flash (high) scores 61.8 and takes 17 s to its first answer token; Gemini 3.8 Flash (medium) scores 60.3 in 11 s, Gemini 3.8 Flash (low) scores 56.0 in 4 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

165 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 5Gemini 3.8 Flash (high)
BenchLeader IndexBenchLeader65.261.8
Agents & tools scoreBenchLeader65.151.0
Coding scoreBenchLeader67.960.2
Human preference scoreBenchLeader75.969.2
Knowledge scoreBenchLeader66.364.1
Maths scoreBenchLeader65.761.8
Reasoning scoreBenchLeader71.263.2
Long context scoreBenchLeader–65.4
Multimodal scoreBenchLeader–67.5
Blended price $/MBenchLeader$10.00$1.50
Output speedBenchLeader63 tok/s112 tok/s
Time to first answerBenchLeader2.8 s16.6 s
Context windowBenchLeader1,000,000 tokens1,048,576 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub92.9%95.4%
Humanity's Last ExamScale AI / CAIS–44.5%†
SimpleBenchSimpleBench80.6%–
Arena Hard PromptsArena (LMArena)1513†1520
LiveBench 2.0 ReasoningLiveBench–89.3%†
LiveBench 1.0 ReasoningLiveBench91.2%†89.3%
GPQA Diamond (AA)Artificial Analysis93.2%†95.3%
Humanity's Last Exam (AA)Artificial Analysis54.9%†47.8%
HLE-Diamond (AA)Artificial Analysis44.8%†33.1%
GPQA Diamond (Vals)Vals AI93.4%94.4%
ARC-AGI-1ARC Prize97.5%†98.5%
ARC-AGI-2ARC Prize90.4%†89.2%
ARC-AGI-3ARC Prize30.2%†35.0%
CritPtArtificial Analysis29.1%†18.3%
CritPt v2 hard subset (AA)Artificial Analysis50.0%†46.5%
MysteryMechanismVals AI37.4%36.5%
Chess PuzzlesEpoch AI Benchmarking Hub43.0%†58.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub37.0%47.0%
LMCAEpoch AI Benchmarking Hub63.3%†51.5%†
DTBenchEpoch AI Benchmarking Hub97.6%†95.7%†
ForecastBenchForecastBench66.6%†–
BTF-3Epoch AI Benchmarking Hub12.0%†–
Coding
SciCodeSciCode56.4%†56.6%
WeirdMLWeirdML86.3%84.8%
FrontierCodeCognition53.4%†41.2%†
Show all 165 benchmarks
MetricPublisherClaude Opus 5Gemini 3.8 Flash (high)
Coding
Arena CodingArena (LMArena)1528†1531
Arena WebDevArena (LMArena)1691†1584
LiveBench 2.0 CodingLiveBench–65.7%†
LiveBench 1.0 CodingLiveBench81.5%†72.5%
LiveCodeBenchVals AI89.0%89.5%
IOIVals AI84.3%56.9%
SWE-bench (Vals)Vals AI97.0%80.0%
Code MigrationVals AI57.5%36.5%
Code Migration v1.1Vals AI–38.6%
ProgramBenchVals AI3.0%1.0%
SWE-bench MultimodalVals AI–40.8%
Vibe Code Bench 1-100Vals AI28.5%18.8%
Vibe Code Bench v1.1Vals AI88.4%78.7%
SWE Atlas: Codebase QnAScale AI SEAL63.2%†47.0%†
SWE Atlas: RefactoringScale AI SEAL–44.8%†
SWE Atlas: Test WritingScale AI SEAL62.2%†53.7%†
Surface Evolver BenchEpoch AI Benchmarking Hub–76.9%
DeepSWE v1.1Epoch AI Benchmarking Hub73.7%†73.8%
CursorBenchCursor46.6%†39.6%
ALE-BenchEpoch AI Benchmarking Hub2164.6†1269.7
FrontierSWEEpoch AI Benchmarking Hub52.0%†19.6%
Terminal-Bench 4.0 (AA)Artificial Analysis49.0%†19.7%
Terminal-Bench 2.1 (AA)Artificial Analysis89.1%†87.6%
Startrise FrontendStartrise82.3%64.6%†
GameBench 2 OverallDuelLab GameBench 276.659.0†
GameBench 2 by reasoning settingDuelLab GameBench 284.7†64.2
Design Arena WebsiteDesign Arena13051296†
FrontierCode 1.1 (AA)Artificial Analysis–36.4%
FrontierCode 2.0 (AA)Artificial Analysis31.1%†18.7%
Agents & tools
Terminal-BenchTerminal-Bench52.1%†18.2%
OSWorld-Verified 2.0OSWorld31.4%†–
Remote Labor IndexScale AI / CAIS–5.8%†
APEX-AgentsMercor65.8%†64.3%†
Arena AgentArena (LMArena)7.5†2.4
LiveBench Long AgenticLiveBench–16.5%†
LiveBench BrowserLiveBench–45.1%†
LiveBench Agentic CodingLiveBench65.2%†54.2%
Terminal-Bench 2.1 (Vals)Vals AI84.6%81.3%
MCP AtlasScale AI SEAL85.8%†–
HiL-BenchScale AI SEAL57.0%41.5%†
τ³-bench Bankingτ²-bench48.7%†–
GDPval-AA v2.1, normalised scoreArtificial Analysis–46.8%
GDPval-AA v2.1Artificial Analysis–1445
τ³-Banking (AA)Artificial Analysis42.1%†45.0%
ITBench SRE (AA)Artificial Analysis–52.5%
Analyst Agent (AA)Artificial Analysis53.8%†46.3%
CUA-benchVals AI9.0%†4.2%
CyberBenchVals AI65.4%43.8%
Finance Agent v2Vals AI58.6%61.4%
Finance Agent v2.1Vals AI–65.8%
Harvey's Legal Agent Benchmark v1.1Vals AI–12.1%
Harvey's Legal Agent Benchmark v1Vals AI6.7%10.0%
Legal Research BenchVals AI55.3%38.9%
SkillsBenchVals AI60.4%58.0%
SREBenchVals AI12.2%11.1%
Tax Agent BenchVals AI47.1%32.1%
Terminal-Bench 4.0 (Vals)Vals AI53.5%19.7%
Terminal-Bench ScienceVals AI27.1%8.6%
Time Horizon Index: KSPVals AI18.8%18.8%
Arena SearchArena (LMArena)1213†–
BALROGBALROG63.4%†–
PostTrainBenchEpoch AI Benchmarking Hub35.0%–
GBAEvalEpoch AI Benchmarking Hub79.6%–
Vending-Bench 2Andon Labs11181.95093.8†
AutomationBenchArtificial Analysis56.6%†59.9%
GDP.pdf (AA)Artificial Analysis21.6%†21.0%
Harvey LAB-AA v1.1Harvey–0.0%
EnterpriseOps-GymArtificial Analysis47.5%†–
AA-Briefcase v1.1Artificial Analysis1655†1203
Design Arena AgenticDesign Arena12631217†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis28.6%†10.0%
CWE-Bench (AA)Artificial Analysis–52.5%
DeepSec-Bench (AA)Artificial Analysis25.9%†14.5%
CyberGym end-to-end (AA)Artificial Analysis–8.4%
FACTS SearchGoogle–80.4%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub85.6%†68.4%
FrontierMath Tier 4Epoch AI Benchmarking Hub73.2%†21.9%
OTIS Mock AIMEEpoch AI Benchmarking Hub97.8%98.9%
ProofBenchVals AI99.0%†48.0%†
LiveBench MathematicsLiveBench95.7%†91.6%
Arena MathsArena (LMArena)1515†1520
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub59.9%†69.7%
LiveBench 2.0 Data AnalysisLiveBench–54.0%†
LiveBench ResearchLiveBench–79.4%†
LiveBench 1.0 Data AnalysisLiveBench74.5%†54.0%
AA-Omniscience v1.1Artificial Analysis37.6†23.3
MMLU-ProVals AI91.6%90.2%
LegalBenchVals AI87.0%87.0%
CorpFinVals AI73.2%–
TaxEvalVals AI75.1%74.5%
PRBench FinanceScale AI SEAL–49.5%†
PRBench LegalScale AI SEAL–48.9%†
BioMysteryBenchVals AI79.3%62.2%
Excel Modeling BenchmarkVals AI73.6%72.2%
Excel Modeling Benchmark v1.1Vals AI–72.9%
MedCodeVals AI63.6%48.1%
MedScribeVals AI91.0%84.5%
Public Benefits Bench v1.1Vals AI76.9%65.3%
DrugDiscoveryBenchScale AI SEAL53.7%†–
EBR-benchEpoch AI Benchmarking Hub45.7%†–
GDP.pdfEpoch AI Benchmarking Hub24.0%†23.2%
AA-Omniscience v1.1: accuracyArtificial Analysis60.9%†50.5%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis40.6%†44.9%
HealthBench Professional (AA)OpenAI–51.7%
Instruction following
LiveBench LanguageLiveBench88.7%†87.8%
LiveBench Instruction FollowingLiveBench63.8%†81.4%
Arena Instruction FollowingArena (LMArena)1494†1492
Human preference
Arena TextArena (LMArena)1489†1498
EQ-Bench 4EQ-Bench1365–
Arena Creative WritingArena (LMArena)1473†1489
Arena Multi-turnArena (LMArena)1486†1505
Arena Longer QueriesArena (LMArena)1502†1511
Multimodal
Arena VisionArena (LMArena)1318†1317
LiveBench DesignLiveBench–69.7%†
MMMU-ProArtificial Analysis84.7%†85.6%
MMMU-Pro (Vals)Vals AI89.9%89.1%
MortgageTaxVals AI72.1%65.3%
SAGEVals AI–88.2%
AudioMCScale AI SEAL–60.4%
AudioMC (text output)Scale AI SEAL–60.4%
Blueprint-Bench 2Epoch AI Benchmarking Hub30.4%38.6%†
Code2Video-AAArtificial Analysis–899
AA-3DArtificial Analysis–974
Long context
AA-LCR v1.1Artificial Analysis79.3%†81.3%
Arena DocumentArena (LMArena)1486†1469
MLCRArtificial Analysis11.1%†10.6%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub164.5158.7†
LiveBench 2.0LiveBench–56.0%†
LiveBench 1.0LiveBench80.1%†75.8%
AA Intelligence Index v5.0Artificial Analysis44.2†30.9
AA Intelligence Index v4.3.2Artificial Analysis50.8†40.9
AA Progress Index v1Artificial Analysis333.7†320.3
Vals IndexVals AI–56.0
Vals Finance IndexVals AI–60.5%
Vals RSI IndexVals AI33.0%†19.9%
Vals Science IndexVals AI47.9%35.8%
Vals Multimodal IndexVals AI73.9%–
Vals Router IndexVals AI–56.0%†
AA Cyber IndexArtificial Analysis–25.1
Safety & honesty
FORTRESSScale AI SEAL16.8%–

Data as of 2026-11-10. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261103-eaea31df94e6 (https://benchmarks.duellab.org/releases/gb2-20261103-eaea31df94e6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • Gemini 3.8 Flash: list price moves to $1.50 / $7.50 per 1M input / output tokens (announced).
  • Claude Opus 5: 1 score revision.
  • Gemini 3.8 Flash (high): 2 score revisions.
  • Gemini 3.8 Flash (high): 1 new result (Arena Document); 1 score revision.
  • Claude Opus 5: 1 score revision.
  • Gemini 3.8 Flash (high): 2 score revisions.
  • Claude Opus 5: 1 score revision.
  • Gemini 3.8 Flash (high): 1 new result (MirrorCode).
  • Claude Opus 5: 1 score revision.
  • Gemini 3.8 Flash (high): 1 score revision.

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

Claude Opus 5 vs Gemini 3.8 Flash: questions

Is Claude Opus 5 better than Gemini 3.8 Flash?
Claude Opus 5 leads on quality: 65.2 vs 61.8. The BenchLeader Index combines every independent quality benchmark; Claude Opus 5 is ahead overall as of 10 Nov 2026, but check the category scores for your use.
Claude Opus 5 vs Gemini 3.8 Flash: which is cheaper, and will that change?
Gemini 3.8 Flash is cheaper today: $1.50 against $10.00 per million tokens, blended at three input tokens per output token. Gemini 3.8 Flash's current price runs until 31 Dec 2026, then $3.00.
Is Claude Opus 5 or Gemini 3.8 Flash better for coding?
Claude Opus 5 scores higher in coding (68 vs 60 on the category index, where 50 is average). On LiveCodeBench, the top shared coding benchmark, it is 89.0% for Claude Opus 5 against 89.5% for Gemini 3.8 Flash.
Which is faster, Claude Opus 5 or Gemini 3.8 Flash?
Gemini 3.8 Flash streams ~1.5× faster (112 vs 63 output tokens per second, Artificial Analysis 7-day median).
Is Claude Opus 5 or Gemini 3.8 Flash better for agents and tool use?
Claude Opus 5 scores higher in agentic tasks (65 vs 51 on the category index, where 50 is average). On Terminal-Bench 2.1 (Vals), the top shared agents & tools benchmark, it is 84.6% for Claude Opus 5 against 81.3% for Gemini 3.8 Flash.
More questions (4)
Which is better for long documents?
Both take about 1M tokens: Claude Opus 5 1,000,000, Gemini 3.8 Flash 1,048,576.
Which has lower latency?
Claude Opus 5 answers first: 2.8 s against 16.6 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.
Do independent benchmarks agree on Claude Opus 5 vs Gemini 3.8 Flash?
Across the 35 benchmarks both have, Claude Opus 5 leads 26, Gemini 3.8 Flash leads 8 and 1 is a tie. Independent sources split: Claude Opus 5 leads on BenchLeader Index and AA Intelligence Index v5.0; Gemini 3.8 Flash leads on Arena Text.
When was Gemini 3.8 Flash released?
Gemini 3.8 Flash was released on 2 Sep 2026.