BenchLeader

Claude Opus 5 vs GPT-6 Astra

Full profiles: Claude Opus 5 (Anthropic) and GPT-6 Astra (OpenAI): every benchmark, price and rank.

Updated · Data from 14 sources

As of 12 Nov 2026, Claude Opus 5 and GPT-6 Astra are neck and neck on quality (64.9 vs 67.1 on the BenchLeader Index, inside the ±3.6 margin), and Claude Opus 5 is 2.0× cheaper at $10.00 vs $20.00 per 1M tokens blended. Choose Claude Opus 5 for human preference (68 vs 66) and multimodal (67 vs 66) and fast first answers (49.8 s vs 236.0 s), where it also streams ~2× faster; choose GPT-6 Astra for knowledge (73 vs 65), reasoning (73 vs 66), and coding (73 vs 68).

Key facts (as of 12 Nov 2026)

  • Claude Opus 5 scores 64.9 on the BenchLeader Index and GPT-6 Astra scores 67.1 (BenchLeader, 12 Nov 2026).
  • Claude Opus 5 costs $10.00 per 1M tokens blended (list price, 12 Nov 2026).
  • GPT-6 Astra costs $20.00 per 1M tokens blended (list price, 12 Nov 2026).
  • Claude Opus 5 streams ~2× faster than GPT-6 Astra (Artificial Analysis, 7-day median, 12 Nov 2026).
  • Of the 61 benchmarks both models have, GPT-6 Astra leads 43 (BenchLeader, 12 Nov 2026).
  • On Arena WebDev, Claude Opus 5 is rated 1691 and GPT-6 Astra 1764 (LMArena, as published by 12 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
CodingGPT-6 Astra73 vs 68 category score; Arena WebDev 1764 vs 1691Medium
Agents & tool useEither64 vs 64 category score; Terminal-Bench 52.1% vs 60.3%Either
MathsGPT-6 Astra71 vs 70 category score; FrontierMath Tiers 1–3 93.7% vs 85.6%Medium
ReasoningGPT-6 Astra73 vs 66 category score; GPQA Diamond 95.8% vs 93.9%Medium
KnowledgeGPT-6 Astra73 vs 65 category score; SimpleQA Verified 75.6% vs 59.9%High
Long documentsEither64 vs 65 category score; AA-LCR v1.1 79.3% vs 80.7%Either
Images and other mediaClaude Opus 567 vs 66 category score; MMMU-Pro 84.7% vs 86.9%Medium
Chat quality (human preference)Claude Opus 568 vs 66 on Arena votesMedium
Lowest costClaude Opus 5$10.00 vs $20.00 per 1M blended; agentic coding session $0.400 vs $0.800–
Fastest first answerClaude Opus 549.8 s (Claude Opus 5 (max)) vs 236.0 s (GPT-6 Astra (max)) 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) and GPT-6 Astra (max) are neck and neck on quality (64.9 vs 67.1, within the ±3.6 margin).
  • Claude Opus 5 (max) is stronger in human preference, multimodal.
  • GPT-6 Astra (max) is stronger in agents & tools, coding, knowledge, long context, maths, reasoning.
  • Claude Opus 5 (max) is 2.0× cheaper ($10.00 vs $20.00 per 1M blended).
  • Claude Opus 5 (max) streams 2.3× faster (80 vs 35 tokens per second).

At a glance

SpecClaude Opus 5GPT-6 Astra
DeveloperAnthropicOpenAI
Released24 Jul 20262 Sep 2026
Context window (tokens)1,000,0001,050,000
Compared atClaude Opus 5 (max)GPT-6 Astra (max)
Open weightsNoNo
Price in / out / cached input per 1M$5.00 / $25.00 / $0.500$10.00 / $50.00 / $1.00

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)GPT-6 Astra (max)Leader
Reasoning66.372.6GPT-6 Astra
Coding67.673.3GPT-6 Astra
Agents & tools63.564.0level
Maths69.870.8GPT-6 Astra
Knowledge65.072.6GPT-6 Astra
Human preference68.166.3Claude Opus 5
Multimodal67.365.7Claude Opus 5
Long context64.365.0level

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 & toolsMathsKnowledgeLong contextMultimodalHumanpreferenceClaude Opus 5, Reasoning: 66.3Claude Opus 5, Coding: 67.6Claude Opus 5, Agents & tools: 63.5Claude Opus 5, Maths: 69.8Claude Opus 5, Knowledge: 65.0Claude Opus 5, Long context: 64.3Claude Opus 5, Multimodal: 67.3Claude Opus 5, Human preference: 68.1GPT-6 Astra, Reasoning: 72.6GPT-6 Astra, Coding: 73.3GPT-6 Astra, Agents & tools: 64.0GPT-6 Astra, Maths: 70.8GPT-6 Astra, Knowledge: 72.6GPT-6 Astra, Long context: 65.0GPT-6 Astra, Multimodal: 65.7GPT-6 Astra, Human preference: 66.3

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

Model, click to add or removeIndex

Where sources disagree

Across the 61 benchmarks both have, Claude Opus 5 leads 17, GPT-6 Astra leads 43 and 1 is a tie. Independent sources split: Claude Opus 5 leads on Arena Text; GPT-6 Astra leads on AA Intelligence Index v5.0.

Overall indexClaude Opus 5 (max)GPT-6 Astra (max)Leader
BenchLeader Index64.967.1neck and neck
AA Intelligence Index v5.044.256.9GPT-6 Astra
Arena Text14891474Claude 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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureClaude Opus 5 (max)GPT-6 Astra (max)
Blended price per 1M tokens$10.00$20.00
Input / output per 1M$5.00 / $25.00$10.00 / $50.00
Cached input per 1M$0.500$1.00
Price per task: chat reply$0.0095$0.019
Price per task: summarize a 30-page report$0.075$0.150
Price per task: code edit$0.068$0.135
Price per task: agentic coding session$0.400$0.800
Price per task: structured extraction$0.015$0.030
Agentic coding session, input mostly cached$0.198$0.395
Cost per benchmark task (AA Index)$11.82$6.24
Output tokens per task (AA Index)118k52k
Of which reasoning73k36k
Time per task (AA Index)24 min25 min
Output speed80 tok/s35 tok/s
Time to first answer49.8 s236.0 s
Time to a complete answer210.6 s550.3 s

Per token is not per task: an average task of Artificial Analysis' Intelligence Index cost $11.82 on Claude Opus 5 (max) and $6.24 on GPT-6 Astra (max), and Claude Opus 5 (max) wrote 2.3× as many output 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 64.9 and takes 50 s to its first answer token; Claude Opus 5 scores 64.7 in 2 s, Claude Opus 5 (high) scores 64.1 in 14 s. GPT-6 Astra (max) scores 67.1 and takes 236 s to its first answer token; GPT-6 Astra scores 66.8 in 5 s, GPT-6 Astra (high) scores 66.3 in 59 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

166 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)GPT-6 Astra (max)
BenchLeader IndexBenchLeader64.967.1
Agents & tools scoreBenchLeader63.564.0
Coding scoreBenchLeader67.673.3
Human preference scoreBenchLeader68.166.3
Knowledge scoreBenchLeader65.072.6
Long context scoreBenchLeader64.365.0
Maths scoreBenchLeader69.870.8
Multimodal scoreBenchLeader67.365.7
Reasoning scoreBenchLeader66.372.6
Blended price $/MBenchLeader$10.00$20.00
Output speedBenchLeader80 tok/s35 tok/s
Time to first answerBenchLeader49.8 s236.0 s
Context windowBenchLeader1,000,000 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.9%95.8%
Humanity's Last ExamScale AI / CAIS–54.8%†
SimpleBenchSimpleBench80.6%†83.6%†
Arena Hard PromptsArena (LMArena)15131500
LiveBench 2.0 ReasoningLiveBench–92.7%†
LiveBench 1.0 ReasoningLiveBench91.2%92.7%
GPQA Diamond (AA)Artificial Analysis93.2%96.1%
Humanity's Last Exam (AA)Artificial Analysis54.9%54.7%
HLE-Diamond (AA)Artificial Analysis44.8%63.8%
GPQA Diamond (Vals)Vals AI93.4%†–
ARC-AGI-1ARC Prize97.5%97.5%
ARC-AGI-2ARC Prize90.4%95.0%
ARC-AGI-3ARC Prize30.2%†98.6%
CritPtArtificial Analysis29.1%31.7%
CritPt v2 hard subset (AA)Artificial Analysis50.0%70.5%
MysteryMechanismVals AI37.4%†53.1%
Chess PuzzlesEpoch AI Benchmarking Hub43.0%74.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub59.0%84.0%
LMCAEpoch AI Benchmarking Hub63.3%64.1%†
DTBenchEpoch AI Benchmarking Hub97.6%97.1%†
ForecastBenchForecastBench66.6%66.6%†
BTF-3Epoch AI Benchmarking Hub12.0%†13.5%†
Coding
SciCodeSciCode56.4%56.5%
WeirdMLWeirdML91.8%93.3%
FrontierCodeCognition53.4%53.3%
Show all 166 benchmarks
MetricPublisherClaude Opus 5 (max)GPT-6 Astra (max)
Coding
GSO-BenchGSO-Bench–79.4%†
Arena CodingArena (LMArena)15281548
Arena WebDevArena (LMArena)16911764
LiveBench 2.0 CodingLiveBench–87.5%†
LiveBench 1.0 CodingLiveBench81.5%80.4%
LiveCodeBenchVals AI89.0%†–
IOIVals AI84.3%†100.0%
SWE-bench (Vals)Vals AI97.0%†–
Code MigrationVals AI57.5%†67.7%
Code Migration v1.1Vals AI–80.6%
ProgramBenchVals AI3.0%†5.5%
SWE-bench MultimodalVals AI–43.3%
Vibe Code Bench 1-100Vals AI28.5%†27.6%
Vibe Code Bench v1.1Vals AI88.4%†89.6%
SWE Atlas: Codebase QnAScale AI SEAL63.2%†59.1%†
SWE Atlas: RefactoringScale AI SEAL–59.0%†
SWE Atlas: Test WritingScale AI SEAL62.2%†50.7%†
MirrorCodeEpoch AI Benchmarking Hub87.8%60.0%
DeepSWE v1.1Epoch AI Benchmarking Hub73.7%73.2%
CursorBenchCursor46.6%–
ALE-BenchEpoch AI Benchmarking Hub2164.6†2951.3
FrontierSWEEpoch AI Benchmarking Hub52.0%65.5%
Terminal-Bench 4.0 (AA)Artificial Analysis49.0%59.1%
Terminal-Bench 2.1 (AA)Artificial Analysis89.1%88.4%
Startrise FrontendStartrise82.3%†79.9%†
GameBench 2 OverallDuelLab GameBench 276.6†88.6†
GameBench 2 by reasoning settingDuelLab GameBench 284.8†94.1†
Design Arena WebsiteDesign Arena1303†1330
FrontierCode 1.1 (AA)Artificial Analysis–47.0%
FrontierCode 2.0 (AA)Artificial Analysis31.1%42.4%
Agents & tools
Terminal-BenchTerminal-Bench52.1%60.3%
OSWorld-Verified 2.0OSWorld31.4%–
Remote Labor IndexScale AI / CAIS–20.8%†
APEX-AgentsMercor65.8%64.7%†
Arena AgentArena (LMArena)7.214.5
LiveBench Long AgenticLiveBench–75.6%†
LiveBench BrowserLiveBench–91.3%†
LiveBench Agentic CodingLiveBench65.2%57.3%
Terminal-Bench 2.1 (Vals)Vals AI84.6%†87.3%
MCP AtlasScale AI SEAL85.8%†–
HiL-BenchScale AI SEAL57.0%†65.3%†
τ³-bench Bankingτ²-bench48.7%–
GDPval-AA v2.1, normalised scoreArtificial Analysis–53.7%
GDPval-AA v2.1Artificial Analysis–1586
τ³-Banking (AA)Artificial Analysis42.1%41.4%
ITBench SRE (AA)Artificial Analysis–48.6%
Analyst Agent (AA)Artificial Analysis53.8%51.3%
CUA-benchVals AI9.0%19.2%
CyberBenchVals AI65.4%†41.1%
Finance Agent v2Vals AI58.6%†53.5%
Finance Agent v2.1Vals AI–60.4%
Harvey's Legal Agent Benchmark v1.1Vals AI–7.1%
Harvey's Legal Agent Benchmark v1Vals AI6.7%†5.4%
Legal Research BenchVals AI55.3%†39.4%
Robo-benchVals AI–23.3%
SkillsBenchVals AI60.4%†–
SREBenchVals AI12.2%†56.9%
Tax Agent BenchVals AI47.1%†20.7%
Terminal-Bench 4.0 (Vals)Vals AI53.5%†59.6%
Terminal-Bench ScienceVals AI27.1%†62.9%
Time Horizon Index: KSPVals AI18.8%†90.5%
Arena SearchArena (LMArena)1213†1245
BALROGBALROG63.4%68.3%
PostTrainBenchEpoch AI Benchmarking Hub35.7%†41.9%
GBAEvalEpoch AI Benchmarking Hub79.6%†–
Vending-Bench 2Andon Labs11181.9†15514.7†
AutomationBenchArtificial Analysis56.6%68.5%
GDP.pdf (AA)Artificial Analysis21.6%31.0%
Harvey LAB-AA v1.1Harvey–8.6%
EnterpriseOps-GymArtificial Analysis47.5%–
AA-Briefcase v1.1Artificial Analysis16551569
GDPxGDPx by SemiAnalysis–9.6%
Design Arena AgenticDesign Arena1264†1385
RoboDojo (LLM as robot policy)RoboDojo–29.0†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis28.6%63.3%
CWE-Bench (AA)Artificial Analysis–63.3%
DeepSec-Bench (AA)Artificial Analysis25.9%36.9%
CyberGym end-to-end (AA)Artificial Analysis–0.0%
FACTS SearchGoogle–93.3%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub85.6%93.7%
FrontierMath Tier 4Epoch AI Benchmarking Hub73.2%97.6%
OTIS Mock AIMEEpoch AI Benchmarking Hub98.9%100.0%
ProofBenchVals AI99.0%99.0%†
LiveBench MathematicsLiveBench95.7%96.8%
Arena MathsArena (LMArena)15151482
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub59.9%75.6%
LiveBench 2.0 Data AnalysisLiveBench–83.0%†
LiveBench ResearchLiveBench–90.6%†
LiveBench 1.0 Data AnalysisLiveBench74.5%83.0%
AA-Omniscience v1.1Artificial Analysis37.642.4
MMLU-ProVals AI91.6%†–
LegalBenchVals AI87.0%†–
CorpFinVals AI73.2%†–
TaxEvalVals AI75.1%†–
PRBench FinanceScale AI SEAL–47.5%†
PRBench LegalScale AI SEAL–48.4%†
BioMysteryBenchVals AI79.3%†79.3%
Excel Modeling BenchmarkVals AI73.6%†71.7%
Excel Modeling Benchmark v1.1Vals AI–72.7%
MedCodeVals AI63.6%†48.5%
MedScribeVals AI91.0%†87.9%
Public Benefits Bench v1.1Vals AI76.9%†–
DrugDiscoveryBenchScale AI SEAL53.7%68.7%
EBR-benchEpoch AI Benchmarking Hub45.7%76.2%
GDP.pdfEpoch AI Benchmarking Hub24.0%34.2%
AA-Omniscience v1.1: accuracyArtificial Analysis60.9%61.6%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis40.6%50.2%
HealthBench Professional (AA)OpenAI–66.0%
Instruction following
LiveBench LanguageLiveBench88.7%89.4%
LiveBench Instruction FollowingLiveBench63.8%75.6%
Arena Instruction FollowingArena (LMArena)14941474
Human preference
Arena TextArena (LMArena)14891474
EQ-Bench 4EQ-Bench1365†–
Arena Creative WritingArena (LMArena)14731450
Arena Multi-turnArena (LMArena)14861483
Arena Longer QueriesArena (LMArena)15021486
Multimodal
Arena VisionArena (LMArena)1318†1279
LiveBench DesignLiveBench–80.0%†
MMMU-ProArtificial Analysis84.7%86.9%
MMMU-Pro (Vals)Vals AI89.9%†–
MortgageTaxVals AI72.1%†–
Blueprint-Bench 2Epoch AI Benchmarking Hub30.4%†49.7%†
Code2Video-AAArtificial Analysis–1427
AA-3DArtificial Analysis–1168
Long context
AA-LCR v1.1Artificial Analysis79.3%80.7%
Arena DocumentArena (LMArena)1486†1474
MLCRArtificial Analysis11.1%41.7%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub164.7†169†
LiveBench 2.0LiveBench–84.5%†
LiveBench 1.0LiveBench80.1%82.2%
AA Intelligence Index v5.0Artificial Analysis44.256.9
AA Intelligence Index v4.3.2Artificial Analysis50.852.7
AA Progress Index v1Artificial Analysis333.7349.6
Vals IndexVals AI–65.0
Vals Finance IndexVals AI–55.6%
Vals RSI IndexVals AI33.0%27.1%
Vals Science IndexVals AI47.9%†65.1%
Vals Multimodal IndexVals AI73.9%†–
Vals Router IndexVals AI–65.0%†
AA Cyber IndexArtificial Analysis–33.4
Safety & honesty
FORTRESSScale AI SEAL16.8%†12.1%†

Data as of 2026-11-12. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261110-2acac640a228 (https://benchmarks.duellab.org/releases/gb2-20261110-2acac640a228/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • GPT-6 Astra (max): 1 new result (PostTrainBench).
  • Claude Opus 5 (max): 1 new result (MirrorCode).
  • GPT-6 Astra (max): 1 new result (MirrorCode); 1 score revision.
  • Claude Opus 5 (max): BenchLeader Index 65.5 → 65.0.
  • GPT-6 Astra (max): BenchLeader Index 68.1 → 67.4.
  • GPT-6 Astra (max): 2 score revisions.
  • GPT-6 Astra (max): 1 score revision.
  • Claude Opus 5 (max): 7 new results (AA Intelligence Index v5.0, HLE-Diamond (AA), AA-Omniscience v1.1…).
  • GPT-6 Astra (max): 1 score revision.
  • GPT-6 Astra (max): 1 score revision.

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

Claude Opus 5 vs GPT-6 Astra: questions

Is Claude Opus 5 better than GPT-6 Astra?
Claude Opus 5 (max) and GPT-6 Astra (max) are neck and neck on quality (64.9 vs 67.1, within the ±3.6 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 12 Nov 2026, so check the category scores for your use.
Claude Opus 5 vs GPT-6 Astra: which is cheaper, and will that change?
Claude Opus 5 is cheaper today: $10.00 against $20.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Claude Opus 5 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher in coding (73 vs 68 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1691 for Claude Opus 5 against 1764 for GPT-6 Astra.
Which is faster, Claude Opus 5 or GPT-6 Astra?
Claude Opus 5 streams ~2× faster (80 vs 35 output tokens per second, Artificial Analysis 7-day median).
Is Claude Opus 5 or GPT-6 Astra better for agents and tool use?
They are level in agentic tasks (64 vs 64 on the category index). On Terminal-Bench, the top shared agents & tools benchmark, it is 52.1% for Claude Opus 5 against 60.3% for GPT-6 Astra.
More questions (4)
Which is better for long documents?
They are level on long-context benchmarks (64 vs 65). Both take about 1.1M tokens: Claude Opus 5 1,000,000, GPT-6 Astra 1,050,000.
Which has lower latency?
Claude Opus 5 answers first: 49.8 s against 236.0 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 GPT-6 Astra?
Across the 61 benchmarks both have, Claude Opus 5 leads 17, GPT-6 Astra leads 43 and 1 is a tie. Independent sources split: Claude Opus 5 leads on Arena Text; GPT-6 Astra leads on AA Intelligence Index v5.0.
When was GPT-6 Astra released?
GPT-6 Astra was released on 2 Sep 2026.