BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

Claude Opus 4.6 vs o3

Full profiles: Claude Opus 4.6 (Anthropic) and o3 (OpenAI): every benchmark, price and rank.

Updated · Data from 18 sources

As of 6 Nov 2026, Claude Opus 4.6 and o3 are neck and neck on quality (62.6 vs 61.0 on the BenchLeader Index, inside the ±4.1 margin), and o3 is 2.9× cheaper at $3.50 vs $10.00 per 1M tokens blended. Choose Claude Opus 4.6 for multimodal (67 vs 57), human preference (70 vs 61), and reasoning (66 vs 61) and fast first answers (1.1 s vs 7.8 s); choose o3 for maths (71 vs 63), where it also streams 3×+ faster.

Key facts (as of 6 Nov 2026)

  • Claude Opus 4.6 scores 62.6 on the BenchLeader Index and o3 scores 61.0 (BenchLeader, 6 Nov 2026).
  • Claude Opus 4.6 costs $10.00 per 1M tokens blended (list price, 6 Nov 2026).
  • o3 costs $3.50 per 1M tokens blended (list price, 6 Nov 2026).
  • o3 streams 3×+ faster than Claude Opus 4.6 (Artificial Analysis, 7-day median, 6 Nov 2026).
  • Of the 8 benchmarks both models have, Claude Opus 4.6 leads 8 (BenchLeader, 6 Nov 2026).
Claude Opus 4.6 benchmarks high Anthropic o3 benchmarks OpenAIAdd a third model

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
CodingEither63 vs 62 category score; Arena Coding 1548 vs 1460Either
Mathso371 vs 63 category score; Arena Maths 1448 vs 1517Medium
ReasoningClaude Opus 4.666 vs 61 category score; Arena Hard Prompts 1531 vs 1441Medium
Images and other mediaClaude Opus 4.667 vs 57 category score; Arena Vision 1316 vs 1214High
Chat quality (human preference)Claude Opus 4.670 vs 61 on Arena votesHigh
Lowest costo3$3.50 vs $10.00 per 1M blended; agentic coding session $0.152 vs $0.400–
Fastest first answerClaude Opus 4.61.1 s (Claude Opus 4.6 (high)) vs 7.8 s (o3) 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 4.6 (high) and o3 are neck and neck on quality (62.6 vs 61.0, within the ±4.1 margin).
  • Claude Opus 4.6 (high) is stronger in coding, human preference, multimodal, reasoning.
  • o3 is stronger in maths, agents & tools, instruction following, knowledge, long context.
  • o3 is 2.9× cheaper ($3.50 vs $10.00 per 1M blended).
  • o3 streams 2.9× faster (110 vs 38 tokens per second).

At a glance

SpecClaude Opus 4.6o3
DeveloperAnthropicOpenAI
Released5 Feb 202616 Apr 2025
Context window (tokens)1,000,000200,000
Compared atClaude Opus 4.6 (high)o3
Open weightsNoNo
Price in / out / cached input per 1M$5.00 / $25.00 / $0.500$2.00 / $8.00 / $0.500

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 4.6 (high)o3Leader
Reasoning65.560.9Claude Opus 4.6
Coding62.561.8level
Agents & tools–70.3not comparable
Maths63.470.5o3
Knowledge–56.1not comparable
Instruction following–65.1not comparable
Human preference69.861.1Claude Opus 4.6
Multimodal66.656.6Claude Opus 4.6
Long context–62.0not 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 avg75ReasoningCodingMathsMultimodalHumanpreferenceClaude Opus 4.6, Reasoning: 65.5Claude Opus 4.6, Coding: 62.5Claude Opus 4.6, Maths: 63.4Claude Opus 4.6, Multimodal: 66.6Claude Opus 4.6, Human preference: 69.8o3, Reasoning: 60.9o3, Coding: 61.8o3, Maths: 70.5o3, Multimodal: 56.6o3, Human preference: 61.1

Not drawn, measured for fewer than two of these models: Agents & tools, Knowledge, Instruction following, Long context.

ModelIndex

Where sources disagree

Across the 8 benchmarks both have, Claude Opus 4.6 leads 8, o3 leads 0. Every overall index that rates both puts Claude Opus 4.6 ahead.

Overall indexClaude Opus 4.6 (high)o3Leader
BenchLeader Index62.661.0neck and neck
Arena Text15031432Claude Opus 4.6

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 4.6 (high)o3
Blended price per 1M tokens$10.00$3.50
Input / output per 1M$5.00 / $25.00$2.00 / $8.00
Cached input per 1M$0.500$0.500
Price per task: chat reply$0.0095$0.0032
Price per task: summarize a 30-page report$0.075$0.029
Price per task: code edit$0.068$0.024
Price per task: agentic coding session$0.400$0.152
Price per task: structured extraction$0.015$0.0056
Agentic coding session, input mostly cached$0.198$0.085
Output speed38 tok/s110 tok/s
Time to first answer1.1 s7.8 s
Time to a complete answer–21.7 s

Each model is compared at its best-ranked reasoning-effort setting. Claude Opus 4.6 (high) scores 62.6 and takes 1 s to its first answer token; Claude Opus 4.6 (thinking) scores 62.4 in 1 s, Claude Opus 4.6 scores 62.3 in 1 s. o3 scores 61.0 and takes 8 s to its first answer token; o3 (high) scores 53.8 in 8 s, o3 (medium) scores 53.3 in 8 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

134 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 4.6 (high)o3
BenchLeader IndexBenchLeader62.661.0
Coding scoreBenchLeader62.561.8
Human preference scoreBenchLeader69.861.1
Maths scoreBenchLeader63.470.5
Multimodal scoreBenchLeader66.656.6
Reasoning scoreBenchLeader65.560.9
Agents & tools scoreBenchLeader–70.3
Instruction following scoreBenchLeader–65.1
Knowledge scoreBenchLeader–56.1
Long context scoreBenchLeader–62.0
Blended price $/MBenchLeader$10.00$3.50
Output speedBenchLeader38 tok/s110 tok/s
Time to first answerBenchLeader1.1 s7.8 s
Context windowBenchLeader1,000,000 tokens200,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub90.5%†81.8%†
Humanity's Last ExamScale AI / CAIS19.0%†20.3%†
SimpleBenchSimpleBench67.6%†53.1%†
Arena Hard PromptsArena (LMArena)15311441
LiveBench 1.0 ReasoningLiveBench88.7%–
GPQA Diamond (AA)Artificial Analysis84.0%†82.7%
Humanity's Last Exam (AA)Artificial Analysis19.1%†20.1%
GPQA Diamond (Vals)Vals AI89.7%†84.1%†
Kagi LLM BenchmarkKagi LLM Benchmark72.4%†67.6%
GPQA Diamond (HELM)HELM Capabilities–75.3%
ARC-AGI-1ARC Prize94.0%60.8%†
ARC-AGI-2ARC Prize69.2%6.5%†
ARC-AGI-3ARC Prize0.5%†–
CritPtArtificial Analysis2.8%†1.1%
SciPredictScale AI SEAL–17.9%
Chess PuzzlesEpoch AI Benchmarking Hub21.0%†36.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub15.0%†29.0%†
CL-benchEpoch AI Benchmarking Hub20.7%†17.8%†
CL-bench LifeEpoch AI Benchmarking Hub17.0%–
LMCAEpoch AI Benchmarking Hub55.8%†39.7%†
DTBenchEpoch AI Benchmarking Hub91.2%†84.8%†
ForecastBenchForecastBench63.4%†64.6%
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub78.7%†62.3%†
WeirdMLWeirdML78.0%52.4%†
FrontierCodeCognition26.6%†–
Show all 134 benchmarks
MetricPublisherClaude Opus 4.6 (high)o3
Coding
GSO-BenchGSO-Bench41.2%8.8%†
Arena CodingArena (LMArena)15481460
Arena WebDevArena (LMArena)1546–
LiveBench 1.0 CodingLiveBench78.2%–
LiveCodeBenchVals AI84.7%†83.9%†
SWE-bench (Vals)Vals AI78.2%†–
SWE-Bench ProScale AI SEAL51.9%†–
Aider PolyglotAider polyglot leaderboard–81.3%
SWE-bench Verified (bash only)SWE-bench75.6%†58.4%
SWE-bench Verified (any scaffold)SWE-bench75.6%†58.4%
Vibe Code Bench v1.1Vals AI57.6%†–
SWE-Bench Pro (private)Scale AI SEAL51.9%†–
SWE Atlas: Codebase QnAScale AI SEAL33.3%†–
SWE Atlas: RefactoringScale AI SEAL35.6%†–
SWE Atlas: Test WritingScale AI SEAL36.7%†–
CadEvalEpoch AI Benchmarking Hub–74.0%†
ALE-BenchEpoch AI Benchmarking Hub996.5†933.5†
AlgoTuneAlgoTune1.5–
GameBench 2 OverallDuelLab GameBench 2–34.6
GameBench 2 by reasoning settingDuelLab GameBench 2–30.5†
LiveCodeBench (AA)Artificial Analysis–80.8%
Agents & tools
Terminal-BenchTerminal-Bench79.8%†–
GDPvalOpenAI–30.8%†
CybenchCybench93.0%†–
Remote Labor IndexScale AI / CAIS4.2%†–
APEX-AgentsMercor46.3%†–
LiveBench Agentic CodingLiveBench49.0%–
Terminal-Bench HardArtificial Analysis48.5%†37.1%
τ²-Bench Telecom (AA)Artificial Analysis84.8%†80.7%
MCP AtlasScale AI SEAL76.8%†–
HiL-BenchScale AI SEAL38.3%†–
BFCL OverallBerkeley Function Calling Leaderboard–63.0%
τ³-bench Bankingτ²-bench27.3%†–
APEX-Agents (AA)Artificial Analysis33.0%†–
Terminal-Bench 2.0 (Vals)Vals AI58.4%†–
DeepResearch BenchDeepResearch Bench55.3%45.2%†
METR Time HorizonsMETR78.9%†63.6%
GBAEvalEpoch AI Benchmarking Hub44.1%†–
Vending-Bench 2Andon Labs8017.6†–
BrowseComp-PlusBrowseComp-Plus–50.5%
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub66.0%†33.3%†
FrontierMath Tier 4Epoch AI Benchmarking Hub26.8%†–
OTIS Mock AIMEEpoch AI Benchmarking Hub94.4%†83.9%†
MATH Level 5Epoch AI Benchmarking Hub–97.8%†
ProofBenchVals AI50.0%†–
LiveBench MathematicsLiveBench89.3%–
AIME (Vals)Vals AI95.6%†85.3%†
MGSMVals AI–91.8%†
AIME 2026MathArena96.7%–
HMMT February 2026MathArena96.2%–
USAMO 2026MathArena47.0%–
IMO 2025MathArena–16.7%†
MathArena ApexMathArena34.5%–
Omni-MATH (HELM)HELM Capabilities–71.4%
MATH 500Vals AI–94.6%†
Arena MathsArena (LMArena)15171448
AIME 2025 (AA)Artificial Analysis–88.3%
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub47.0%†49.4%†
LiveBench 1.0 Data AnalysisLiveBench69.9%–
MMLU-ProVals AI89.1%†85.6%†
LegalBenchVals AI85.3%†83.8%†
CorpFinVals AI67.0%†59.7%†
TaxEvalVals AI76.0%†74.6%†
MedQAVals AI95.4%†96.1%†
PRBench FinanceScale AI SEAL53.3%†47.7%
PRBench LegalScale AI SEAL52.3%†48.6%
MultiNRCScale AI SEAL48.3%†45.5%†
MMLU-Pro (HELM)HELM Capabilities–85.9%
CaseLaw v2Vals AI62.1%†–
MedCodeVals AI48.2%†47.3%†
MedScribeVals AI86.7%†76.7%†
DrugDiscoveryBenchScale AI SEAL27.7%†–
EBR-benchEpoch AI Benchmarking Hub12.7%†–
Global-MMLU-Lite (AA)Artificial Analysis92.2%†–
Instruction following
LiveBench LanguageLiveBench83.3%–
LiveBench Instruction FollowingLiveBench63.3%–
IFBenchArtificial Analysis44.6%†71.4%
MultiChallengeScale AI SEAL56.0%†56.6%†
TutorBenchScale AI SEAL53.5%†52.1%†
IFEval (HELM)HELM Capabilities–86.9%
WildBench (HELM)HELM Capabilities–86.1%
Arena Instruction FollowingArena (LMArena)15111403
Human preference
Arena TextArena (LMArena)15031432
EQ-Bench 4EQ-Bench1210†–
Arena Creative WritingArena (LMArena)14991383
Arena Multi-turnArena (LMArena)15171420
Arena Longer QueriesArena (LMArena)15211411
Lech Mazur Creative WritingLech Mazur–83.9%†
Multimodal
Arena VisionArena (LMArena)13161214
MMMU-ProArtificial Analysis72.5%†70.1%
VISTAScale AI SEAL45.5%†50.1%†
MMMU (validation)MMMU–82.9%
MMMU-Pro (official)MMMU–76.4%
MMMU-Pro (Vals)Vals AI83.9%†80.4%†
MortgageTaxVals AI68.5%†65.7%†
VTBScale AI SEAL27.5%†13.7%
GeoBenchGeoBench–60.0%†
VPCTEpoch AI Benchmarking Hub–52.0%†
Long context
Fiction.LiveBench 120kFiction.live–100.0%†
AA-LCR v1.1Artificial Analysis67.0%†74.7%
Arena DocumentArena (LMArena)1495–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub156.3†147.0
LiveBench 1.0LiveBench74.5%–
AA Intelligence Index v4.3.2Artificial Analysis26.4†20.2
AA Progress Index v1Artificial Analysis274.5†258.3
HELM Capabilities meanHELM Capabilities–81.1%
Safety & honesty
FORTRESSScale AI SEAL20.5%†16.0%
MASKScale AI SEAL96.3%†84.5%†
PropensityBenchScale AI SEAL–10.5%

Data as of 2026-11-06. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261103-09b2218cbd85 (https://benchmarks.duellab.org/releases/gb2-20261103-09b2218cbd85/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • o3: 1 score revision.
  • o3: 1 new result (AA Progress Index v1).
  • o3: 1 score revision.
  • Claude Opus 4.6 (high): 1 new result (USAMO 2026).
  • o3: 1 new result (PRBench Legal).
  • o3: 1 new result (VTB).
  • o3: 1 new result (PRBench Legal).
  • o3: 1 new result (PRBench Finance).
  • o3: 1 new result (SciPredict).
  • Claude Opus 4.6 (high): BenchLeader Index 61.3 → 63.0.

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

Claude Opus 4.6 vs o3: questions

Is Claude Opus 4.6 better than o3?
Claude Opus 4.6 (high) and o3 are neck and neck on quality (62.6 vs 61.0, within the ±4.1 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 6 Nov 2026, so check the category scores for your use.
Claude Opus 4.6 vs o3: which is cheaper, and will that change?
o3 is cheaper today: $3.50 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 4.6 or o3 better for coding?
They are level in coding (63 vs 62 on the category index). On Arena Coding, the top shared coding benchmark, it is 1548 for Claude Opus 4.6 against 1460 for o3.
Which is faster, Claude Opus 4.6 or o3?
o3 streams 3×+ faster (110 vs 38 output tokens per second, Artificial Analysis 7-day median).
Which is better for long documents?
Claude Opus 4.6 accepts more context: 1,000,000 against 200,000 tokens for o3.
More questions (3)
Which has lower latency?
Claude Opus 4.6 answers first: 1.1 s against 7.8 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 4.6 vs o3?
Across the 8 benchmarks both have, Claude Opus 4.6 leads 8, o3 leads 0. Every overall index that rates both puts Claude Opus 4.6 ahead.
When was Claude Opus 4.6 released?
Claude Opus 4.6 was released on 5 Feb 2026.