BenchLeader

Claude Opus 5.5 vs GPT-6.2 Astra

Full profiles: Claude Opus 5.5 (Anthropic) and GPT-6.2 Astra (OpenAI): every benchmark, price and rank.

Updated · Data from 13 sources

As of 19 Nov 2026, Claude Opus 5.5 and GPT-6.2 Astra are neck and neck on quality (67.8 vs 68.9 on the BenchLeader Index, inside the ±3.9 margin), and Claude Opus 5.5 is 2.0× cheaper at $8.00 vs $16.00 per 1M tokens blended. Choose Claude Opus 5.5 for coding (77 vs 76), where it also streams ~2× faster; choose GPT-6.2 Astra for fast first answers (202.7 s vs 352.2 s) and knowledge (74 vs 71), agents & tools (66 vs 64), and reasoning (73 vs 71).

Key facts (as of 19 Nov 2026)

  • Claude Opus 5.5 scores 67.8 on the BenchLeader Index and GPT-6.2 Astra scores 68.9 (BenchLeader, 19 Nov 2026).
  • Claude Opus 5.5 costs $8.00 per 1M tokens blended (list price, 19 Nov 2026).
  • GPT-6.2 Astra costs $16.00 per 1M tokens blended (list price, 19 Nov 2026).
  • Claude Opus 5.5 streams ~2× faster than GPT-6.2 Astra (Artificial Analysis, 7-day median, 19 Nov 2026).
  • Of the 64 benchmarks both models have, GPT-6.2 Astra leads 42 (BenchLeader, 19 Nov 2026).
  • On Arena WebDev, Claude Opus 5.5 is rated 1818 and GPT-6.2 Astra 1781 (LMArena, as published by 19 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 5.577 vs 76 category score; Arena WebDev 1818 vs 1781Medium
Agents & tool useGPT-6.2 Astra66 vs 64 category score; CWE-Bench (AA) 65.0% vs 58.3%Medium
MathsEither73 vs 73 category score; FrontierMath Tiers 1–3 91.2% vs 92.6%Either
ReasoningGPT-6.2 Astra73 vs 71 category score; GPQA Diamond 96.8% vs 90.6%Medium
KnowledgeGPT-6.2 Astra74 vs 71 category score; SimpleQA Verified 78.5% vs 72.2%Medium
Images and other mediaGPT-6.2 Astra72 vs 70 category score; MMMU-Pro 89.5% vs 87.7%Medium
Lowest costClaude Opus 5.5$8.00 vs $16.00 per 1M blended; agentic coding session $0.320 vs $0.640–
Fastest first answerGPT-6.2 Astra202.7 s (GPT-6.2 Astra (max)) vs 352.2 s (Claude Opus 5.5 (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 5.5 (max) and GPT-6.2 Astra (max) are neck and neck on quality (67.8 vs 68.9, within the ±3.9 margin).
  • Claude Opus 5.5 (max) is stronger in coding, maths.
  • GPT-6.2 Astra (max) is stronger in agents & tools, composite, knowledge, multimodal, reasoning.
  • Claude Opus 5.5 (max) is 2.0× cheaper ($8.00 vs $16.00 per 1M blended).
  • Claude Opus 5.5 (max) streams 2.4× faster (115 vs 49 tokens per second).

At a glance

SpecClaude Opus 5.5GPT-6.2 Astra
DeveloperAnthropicOpenAI
Released22 Sep 202617 Nov 2026
Context window (tokens)1,000,0001,050,000
Max output (tokens)128,000 (up to 300K on the Message Batches API (beta))–
Input typestext, image–
Reasoning effort settingslow, medium, high, xhigh, max–
Compared atClaude Opus 5.5 (max)GPT-6.2 Astra (max)
Open weightsNoNo
Price in / out / cached input per 1M$4.00 / $20.00 / $0.200$8.00 / $40.00 / $0.400
API model IDclaude-opus-5-5–
Where to use itClaude API, Amazon Bedrock, Google Cloud, Microsoft Foundry–

Limits, availability and announced prices from the vendors' own pages: platform.claude.com (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 5.5 (max)GPT-6.2 Astra (max)Leader
Reasoning70.773.1GPT-6.2 Astra
Coding77.476.1Claude Opus 5.5
Agents & tools64.066.4GPT-6.2 Astra
Maths72.772.5level
Knowledge71.274.2GPT-6.2 Astra
Multimodal69.971.7GPT-6.2 Astra
Long context67.1–not comparable
Composite68.069.5GPT-6.2 Astra

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 & toolsMathsKnowledgeMultimodalClaude Opus 5.5, Reasoning: 70.7Claude Opus 5.5, Coding: 77.4Claude Opus 5.5, Agents & tools: 64.0Claude Opus 5.5, Maths: 72.7Claude Opus 5.5, Knowledge: 71.2Claude Opus 5.5, Multimodal: 69.9GPT-6.2 Astra, Reasoning: 73.1GPT-6.2 Astra, Coding: 76.1GPT-6.2 Astra, Agents & tools: 66.4GPT-6.2 Astra, Maths: 72.5GPT-6.2 Astra, Knowledge: 74.2GPT-6.2 Astra, Multimodal: 71.7

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

Model, click to add or removeIndex

Where sources disagree

Across the 64 benchmarks both have, Claude Opus 5.5 leads 21, GPT-6.2 Astra leads 42 and 1 is a tie. Independent sources split: Claude Opus 5.5 leads on Vals Index; GPT-6.2 Astra leads on AA Intelligence Index v5.0.1.

Overall indexClaude Opus 5.5 (max)GPT-6.2 Astra (max)Leader
BenchLeader Index67.868.9neck and neck
AA Intelligence Index v5.0.159.162.1GPT-6.2 Astra
Vals Index70.167.7Claude Opus 5.5

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 5.5 (max)GPT-6.2 Astra (max)
Blended price per 1M tokens$8.00$16.00
Input / output per 1M$4.00 / $20.00$8.00 / $40.00
Cached input per 1M$0.200$0.400
Price per task: chat reply$0.0076$0.015
Price per task: summarize a 30-page report$0.060$0.120
Price per task: code edit$0.054$0.108
Price per task: agentic coding session$0.320$0.640
Price per task: structured extraction$0.012$0.024
Agentic coding session, input mostly cached$0.149$0.298
Cost per benchmark task (AA Index)$10.04$4.00
Output tokens per task (AA Index)173k41k
Of which reasoning120k26k
Time per task (AA Index)25 min14 min
Output speed115 tok/s49 tok/s
Time to first answer352.2 s202.7 s
Time to a complete answer, last measured 13 Nov 2026711.4 s–

Per token is not per task: an average task of Artificial Analysis' Intelligence Index cost $10.04 on Claude Opus 5.5 (max) and $4.00 on GPT-6.2 Astra (max), and Claude Opus 5.5 (max) wrote 4.2× 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 5.5 (max) scores 67.8 and takes 352 s to its first answer token; Claude Opus 5.5 (high) scores 65.5 in 22 s, Claude Opus 5.5 (xhigh) scores 64.6 in 64 s. GPT-6.2 Astra (max) scores 68.9 and takes 203 s to its first answer token; GPT-6.2 Astra (xhigh) scores 67.1 in 67 s, GPT-6.2 Astra (medium) scores 65.6 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

169 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 5.5 (max)GPT-6.2 Astra (max)
BenchLeader IndexBenchLeader67.868.9
Agents & tools scoreBenchLeader64.066.4
Coding scoreBenchLeader77.476.1
Composite scoreBenchLeader68.069.5
Knowledge scoreBenchLeader71.274.2
Long context scoreBenchLeader67.1–
Maths scoreBenchLeader72.772.5
Multimodal scoreBenchLeader69.971.7
Reasoning scoreBenchLeader70.773.1
Blended price $/MBenchLeader$8.00$16.00
Output speedBenchLeader115 tok/s49 tok/s
Time to first answerBenchLeader352.2 s202.7 s
Context windowBenchLeader1,000,000 tokens1,050,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub90.6%96.8%
SimpleBenchSimpleBench88.4%†–
Arena Hard PromptsArena (LMArena)1531†–
LiveBench 2.0 ReasoningLiveBench92.2%91.6%
LiveBench 1.0 ReasoningLiveBench92.2%–
Humanity's Last Exam (AA)Artificial Analysis61.4%–
HLE-Diamond (AA)Artificial Analysis67.9%72.2%
ARC-AGI-1ARC Prize97.5%98.5%
ARC-AGI-2ARC Prize91.7%95.8%
ARC-AGI-3ARC Prize–99.8%
CritPtArtificial Analysis31.7%–
CritPt v2 hard subset (AA)Artificial Analysis68.0%71.5%
MysteryMechanismVals AI49.5%53.1%
HLE-Diamond (Scale)Scale AI SEAL55.0%†66.9%†
Arena ExpertArena (LMArena)1550†–
Chess PuzzlesEpoch AI Benchmarking Hub67.0%78.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub71.0%95.0%
LMCAEpoch AI Benchmarking Hub68.2%–
DTBenchEpoch AI Benchmarking Hub98.9%–
ForecastBenchForecastBench66.6%†–
Coding
SciCodeSciCode66.9%–
FrontierCodeCognition54.6%†–
Arena CodingArena (LMArena)1547†–
Arena WebDevArena (LMArena)18181781
LiveBench 2.0 CodingLiveBench87.1%91.9%
Show all 169 benchmarks
MetricPublisherClaude Opus 5.5 (max)GPT-6.2 Astra (max)
Coding
LiveBench 1.0 CodingLiveBench89.3%–
IOIVals AI95.1%–
Code MigrationVals AI66.7%–
Code Migration v1.1Vals AI81.0%–
ProgramBenchVals AI18.5%–
SWE-bench MultimodalVals AI70.2%–
Vibe Code Bench 1-100Vals AI30.4%†–
Vibe Code Bench v1.1Vals AI90.3%89.9%
Vibe Code Bench 1-100 v1.1Vals AI30.9%55.2%
SWE Atlas v1.1Scale AI SEAL77.5%†60.5%†
SWE Atlas v1.1: Codebase QnAScale AI SEAL77.4%†67.7%†
SWE Atlas v1.1: Test WritingScale AI SEAL77.8%†62.2%†
SWE Atlas v1.1: RefactoringScale AI SEAL77.1%†51.4%†
MirrorCodeEpoch AI Benchmarking Hub95.6%67.8%
CursorBenchCursor57.8%–
ALE-BenchEpoch AI Benchmarking Hub2147.4†–
FrontierSWEEpoch AI Benchmarking Hub62.3%–
WeirdML v3WeirdML31.2%†–
Terminal-Bench 4.0 (AA)Artificial Analysis59.6%64.1%
Startrise FrontendStartrise80.2%†–
GameBench 2 OverallDuelLab GameBench 287.6†–
GameBench 2 by reasoning settingDuelLab GameBench 294.7†–
Design Arena WebsiteDesign Arena1321†1339†
FrontierCode 1.1 (AA)Artificial Analysis52.2%–
FrontierCode 2.0 (AA)Artificial Analysis46.4%45.4%
Agents & tools
Terminal-BenchTerminal-Bench65.8%62.1%
Remote Labor IndexScale AI / CAIS21.3%†–
APEX-AgentsMercor73.5%–
Arena AgentArena (LMArena)18.615.2
LiveBench Long AgenticLiveBench72.8%70.8%
LiveBench BrowserLiveBench88.4%93.0%
LiveBench Agentic CodingLiveBench71.7%–
Terminal-Bench 2.1 (Vals)Vals AI87.6%†–
MCP AtlasScale AI SEAL85.4%†–
HiL-BenchScale AI SEAL59.0%†–
GDPval-AA v2.1, normalised scoreArtificial Analysis68.3%–
GDPval-AA v2.1Artificial Analysis1866–
ITBench SRE (AA)Artificial Analysis38.2%–
Analyst Agent (AA)Artificial Analysis56.3%62.5%
CUA-benchVals AI14.0%–
CyberBenchVals AI55.4%–
Finance Agent v2Vals AI58.6%–
Finance Agent v2.1Vals AI67.8%60.8%
Harvey's Legal Agent Benchmark v1.1Vals AI9.2%16.3%
Harvey's Legal Agent Benchmark v1Vals AI3.8%–
Legal Research BenchVals AI50.5%42.8%
Robo-benchVals AI14.6%32.7%
SREBenchVals AI33.6%–
Tax Agent BenchVals AI44.5%21.8%
Terminal-Bench 4.0 (Vals)Vals AI65.2%63.1%
Terminal-Bench ScienceVals AI47.1%71.4%
Time Horizon Index: KSPVals AI91.3%–
Web Search IndexVals AI53.8%–
Arena SearchArena (LMArena)1213†–
Arena Agent Task CompletionArena (LMArena)19.212.4
PostTrainBenchEpoch AI Benchmarking Hub43.8%–
GBAEvalEpoch AI Benchmarking Hub90.6%†–
Vending-Bench 2Andon Labs9235.3†–
SpireClimbEpoch AI Benchmarking Hub5.8%41.3%
AutomationBenchArtificial Analysis69.5%77.3%
GDP.pdf (AA)Artificial Analysis26.2%29.8%
Harvey LAB-AA v1.1Harvey4.2%15.0%
EnterpriseOps-GymArtificial Analysis–29.9%
AA-Briefcase v1.1Artificial Analysis18001643
GDPxGDPx by SemiAnalysis11.9%–
Design Arena AgenticDesign Arena1325†1304
Terminal-Bench-Science 0.1 (AA)Artificial Analysis59.0%78.1%
CWE-Bench (AA)Artificial Analysis58.3%65.0%
DeepSec-Bench (AA)Artificial Analysis26.9%46.8%
CyberGym end-to-end (AA)Artificial Analysis0.8%0.0%
MKA (AA)Artificial Analysis53.0%52.7%
FACTS SearchGoogle88.4%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub91.2%92.6%
FrontierMath Tier 4Epoch AI Benchmarking Hub95.0%100.0%
OTIS Mock AIMEEpoch AI Benchmarking Hub100.0%100.0%
ProofBenchVals AI100.0%–
LiveBench MathematicsLiveBench97.1%–
ArXivMathMathArena82.6%†–
BrokenArXivMathArena77.3%†–
Autoform-HardVals AI–87.5%
Arena MathsArena (LMArena)1516†–
FrontierMath ErdősEpoch AI Benchmarking Hub2.9%10.3%
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub72.2%78.5%
LiveBench 2.0 Data AnalysisLiveBench80.3%84.1%
LiveBench ResearchLiveBench88.9%91.5%
LiveBench 1.0 Data AnalysisLiveBench80.3%–
AA-Omniscience v1.1Artificial Analysis48.150
PRBench LegalScale AI SEAL47.6%†43.1%†
BioMysteryBenchVals AI79.3%80.7%
ClaimBenchVals AI100.0%–
Excel Modeling BenchmarkVals AI75.9%–
Excel Modeling Benchmark v1.1Vals AI75.8%76.0%
MedCodeVals AI49.8%–
MedScribeVals AI91.4%–
Public Benefits Bench v1.1Vals AI70.6%–
Access to Justice BenchmarkVals AI73.8%–
Financial Coaching BenchVals AI76.2%–
Immigration BenchVals AI72.1%–
DrugDiscoveryBenchScale AI SEAL–65.5%†
EBR-benchEpoch AI Benchmarking Hub71.4%92.4%
GDP.pdfEpoch AI Benchmarking Hub48.8%–
AA-Omniscience v1.1: accuracyArtificial Analysis65.5%68.6%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis49.7%40.9%
HealthBench Professional (AA)OpenAI59.9%–
FACTS ParametricGoogle58.7%†–
Instruction following
LiveBench LanguageLiveBench86.3%–
LiveBench Instruction FollowingLiveBench65.7%–
MultiChallengeScale AI SEAL–73.2%†
Arena Instruction FollowingArena (LMArena)1517†–
Human preference
Arena TextArena (LMArena)1507†–
EQ-Bench 4EQ-Bench1340†–
EQ-Bench Creative Writing v3EQ-Bench2051†–
Arena Creative WritingArena (LMArena)1516†–
Arena Multi-turnArena (LMArena)1501†–
Arena Longer QueriesArena (LMArena)1522†–
Multimodal
Arena VisionArena (LMArena)1324†–
LiveBench DesignLiveBench83.3%82.8%
MMMU-ProArtificial Analysis87.7%89.5%
SAGEVals AI91.3%–
Humanity's Sixth SenseScale AI SEAL44.6%†63.2%†
Blueprint-Bench 2Epoch AI Benchmarking Hub51.2%†–
Furniture AssemblyEpoch AI Benchmarking Hub83.3%89.2%
VIALS-AAHandshake AI61.5%65.8%
Code2Video-AAArtificial Analysis1536–
AA-3DArtificial Analysis12021318
FACTS MultimodalGoogle56.4%†–
Long context
AA-LCR v1.1Artificial Analysis84.7%–
MLCRArtificial Analysis27.8%39.4%
FACTS GroundingGoogle62.5%†–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub168.7†174.2†
LiveBench 2.0LiveBench83.4%84.9%
LiveBench 1.0LiveBench83.2%–
AA Intelligence Index v5.0.1Artificial Analysis59.162.1
AA Intelligence Index v4.3.2Artificial Analysis57.6–
AA Progress Index v1Artificial Analysis350–
Vals IndexVals AI70.167.7
Vals Finance IndexVals AI65.3%57.3%
Vals RSI IndexVals AI37.3%29.0%
Vals Science IndexVals AI58.6%68.4%
Vals Router IndexVals AI64.2%†–
Vals Legal IndexVals AI72.4%69.7%
AA Cyber IndexArtificial Analysis28.737.3
Safety & honesty
FORTRESSScale AI SEAL5.8%–
DistressBenchScale AI SEAL75.6%†–

Data as of 2026-11-19. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261114-ca113f475f5e (https://benchmarks.duellab.org/releases/gb2-20261114-ca113f475f5e/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • GPT-6.2 Astra (max): BenchLeader Index 67.8 → 68.9.
  • GPT-6.2 Astra (max): 6 new results (VIALS-AA, Arena WebDev, Arena Agent…).
  • GPT-6.2 Astra (max): BenchLeader Index 67.2 → 67.8.
  • GPT-6.2 Astra (max): 2 new results (MKA (AA), Terminal-Bench).
  • Claude Opus 5.5 (max): BenchLeader Index 68.3 → 67.6.
  • Claude Opus 5.5 (max): 1 score revision.
  • Claude Opus 5.5 (max): BenchLeader Index 68.8 → 68.3.
  • Claude Opus 5.5 (max): 2 score revisions.
  • Claude Opus 5.5 (max): 38 new results (LiveBench 2.0, LiveBench Long Agentic, LiveBench 2.0 Reasoning…).
  • Claude Opus 5.5 (max): 1 new result (PostTrainBench); 1 score revision.

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

Claude Opus 5.5 vs GPT-6.2 Astra: questions

Is Claude Opus 5.5 better than GPT-6.2 Astra?
Claude Opus 5.5 (max) and GPT-6.2 Astra (max) are neck and neck on quality (67.8 vs 68.9, within the ±3.9 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 19 Nov 2026, so check the category scores for your use.
Claude Opus 5.5 vs GPT-6.2 Astra: which is cheaper, and will that change?
Claude Opus 5.5 is cheaper today: $8.00 against $16.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Claude Opus 5.5 or GPT-6.2 Astra better for coding?
Claude Opus 5.5 scores higher in coding (77 vs 76 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1818 for Claude Opus 5.5 against 1781 for GPT-6.2 Astra.
Which is faster, Claude Opus 5.5 or GPT-6.2 Astra?
Claude Opus 5.5 streams ~2× faster (115 vs 49 output tokens per second, Artificial Analysis 7-day median).
Is Claude Opus 5.5 or GPT-6.2 Astra better for agents and tool use?
GPT-6.2 Astra scores higher in agentic tasks (66 vs 64 on the category index, where 50 is average). On Terminal-Bench, the top shared agents & tools benchmark, it is 65.8% for Claude Opus 5.5 against 62.1% for GPT-6.2 Astra.
More questions (4)
Which is better for long documents?
Both take about 1.1M tokens: Claude Opus 5.5 1,000,000, GPT-6.2 Astra 1,050,000.
Which has lower latency?
GPT-6.2 Astra answers first: 202.7 s against 352.2 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 5.5 vs GPT-6.2 Astra?
Across the 64 benchmarks both have, Claude Opus 5.5 leads 21, GPT-6.2 Astra leads 42 and 1 is a tie. Independent sources split: Claude Opus 5.5 leads on Vals Index; GPT-6.2 Astra leads on AA Intelligence Index v5.0.1.
When was GPT-6.2 Astra released?
GPT-6.2 Astra was released on 17 Nov 2026.