BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

Claude Opus 5 vs Grok 4.6

Full profiles: Claude Opus 5 (Anthropic) and Grok 4.6 (SpaceXAI): every benchmark, price and rank.

Updated · Data from 11 sources

As of 25 Oct 2026, Claude Opus 5 (max) and Grok 4.6 (medium) are neck and neck on quality (68.8 vs 64.2 on the BenchLeader Index, inside the ±5.1 margin of error), and Grok 4.6 (medium) costs 3.3× less, which makes it the better buy for most workloads. Pick Claude Opus 5 for reasoning (74 vs 64); pick Grok 4.6 for agents & tools (74 vs 64).

Claude Opus 5 benchmarks max Anthropic Grok 4.6 benchmarks medium SpaceXAIOpen 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 62 category score; SciCode 56.4% vs 55.9%High
Agents & tool useGrok 4.674 vs 64 category score; τ³-Banking (AA) 44.3% vs 42.1%High
ReasoningClaude Opus 574 vs 64 category score; ARC-AGI-2 90.4% vs 61.3%High
KnowledgeGrok 4.674 vs 72 category score; AA-Omniscience 28 vs 37.1Medium
Long documentsEither65 vs 66 category score; AA-LCR 79.3% vs 81.0%Either
Chat quality (human preference)–Only Claude Opus 5 has an Arena rating (69 category score)Not comparable
Lowest costGrok 4.6$3.00 vs $10.00 per 1M blended; agentic coding session $0.144 vs $0.400–
Fastest first answerEither24.1 s (Claude Opus 5 (max)) vs 25.5 s (Grok 4.6 (medium)) 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 Grok 4.6 (medium) are neck and neck on quality (68.8 vs 64.2, within the ±5.1 margin).
  • Claude Opus 5 (max) is stronger in coding, composite, human preference, maths, multimodal, reasoning.
  • Grok 4.6 (medium) is stronger in agents & tools, knowledge, long context.
  • Grok 4.6 (medium) is 3.3× cheaper ($3.00 vs $10.00 per 1M blended).
  • Grok 4.6 (medium) streams 1.4× faster (78 vs 57 tokens per second).

At a glance

SpecClaude Opus 5Grok 4.6
DeveloperAnthropicSpaceXAI
Released24 Jul 202611 Aug 2026
Context window (tokens)1,000,000500,000
Compared atClaude Opus 5 (max)Grok 4.6 (medium)
Open weightsNoNo
Price in / out / cached input per 1M$5.00 / $25.00 / $0.500$2.00 / $6.00 / $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 5 (max)Grok 4.6 (medium)Leader
Reasoning74.164.4Claude Opus 5
Coding70.262.3Claude Opus 5
Agents & tools64.373.5Grok 4.6
Maths70.6–not comparable
Knowledge71.674.1Grok 4.6
Human preference68.5–not comparable
Multimodal67.9–not comparable
Long context64.765.6level
Composite86.377.4Claude Opus 5

Where sources disagree

Across the 18 benchmarks both have, Claude Opus 5 leads 12, Grok 4.6 leads 6. Every overall index that rates both puts Claude Opus 5 ahead.

Overall indexClaude Opus 5 (max)Grok 4.6 (medium)Leader
BenchLeader Index68.864.2neck and neck
AA Intelligence Index v4.3.250.842.8Claude Opus 5
Arena Text14911456†Claude 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)Grok 4.6 (medium)
Blended price per 1M tokens$10.00$3.00
Input / output per 1M$5.00 / $25.00$2.00 / $6.00
Cached input per 1M$0.500$0.500
Price per task: chat reply$0.0095$0.0026
Price per task: summarize a 30-page report$0.075$0.028
Price per task: code edit$0.068$0.021
Price per task: agentic coding session$0.400$0.144
Price per task: structured extraction$0.015$0.0052
Cost to run a full benchmark suite$5.86$1.14
Output speed57 tok/s78 tok/s
Time to first answer24.1 s25.5 s
Time to a complete answer32.9 s31.9 s

Per token is not per task: on Artificial Analysis' full suite Claude Opus 5 (max) cost $5.86 and Grok 4.6 (medium) $1.14, because Claude Opus 5 (max) used about 1.5× 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. Grok 4.6 (medium) scores 64.2 and takes 25 s to its first answer token; Grok 4.6 (high) scores 62.9 in 25 s, Grok 4.6 (xhigh) scores 62.5 in 29 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

129 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)Grok 4.6 (medium)
BenchLeader IndexBenchLeader68.864.2
Agents & tools scoreBenchLeader64.373.5
Coding scoreBenchLeader70.262.3
Composite scoreBenchLeader86.377.4
Human preference scoreBenchLeader68.5–
Knowledge scoreBenchLeader71.674.1
Long context scoreBenchLeader64.765.6
Maths scoreBenchLeader70.6–
Multimodal scoreBenchLeader67.9–
Reasoning scoreBenchLeader74.164.4
Blended price $/MBenchLeader$10.00$3.00
Output speedBenchLeader57 tok/s78 tok/s
Time to first answerBenchLeader24.1 s25.5 s
Context windowBenchLeader1,000,000 tokens500,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.9%94.0%†
SimpleBenchSimpleBench80.6%†75.9%†
Arena Hard PromptsArena (LMArena)15151482†
LiveBench 2.0 ReasoningLiveBench91.2%†91.2%†
LiveBench 1.0 ReasoningLiveBench91.2%90.5%†
GPQA Diamond (AA)Artificial Analysis93.2%93.5%
Humanity's Last Exam (AA)Artificial Analysis54.9%42.1%
GPQA Diamond (Vals)Vals AI93.4%†94.7%†
ARC-AGI-1ARC Prize97.5%87.5%
ARC-AGI-2ARC Prize90.4%61.3%
ARC-AGI-3ARC Prize30.2%†2.1%†
CritPtArtificial Analysis29.1%17.7%
MysteryMechanismVals AI37.4%†30.6%†
Chess PuzzlesEpoch AI Benchmarking Hub43.0%33.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub59.0%34.0%†
LMCAEpoch AI Benchmarking Hub63.3%48.5%†
DTBenchEpoch AI Benchmarking Hub97.6%97.3%†
ForecastBenchForecastBench66.5%65.5%†
BTF-3Epoch AI Benchmarking Hub12.0%†–
Coding
SciCodeSciCode56.4%55.9%
WeirdMLWeirdML91.8%67.3%†
FrontierCodeCognition53.4%48.0%†
Arena CodingArena (LMArena)15281510†
Arena WebDevArena (LMArena)16911617†
LiveBench 2.0 CodingLiveBench90.8%†89.1%†
Show all 129 benchmarks
MetricPublisherClaude Opus 5 (max)Grok 4.6 (medium)
Coding
LiveBench 1.0 CodingLiveBench81.5%76.8%†
LiveCodeBenchVals AI89.0%†88.2%†
IOIVals AI84.3%†47.6%†
SWE-bench (Vals)Vals AI97.0%†95.6%†
Code MigrationVals AI57.5%†44.6%†
ProgramBenchVals AI3.0%†1.0%†
Vibe Code Bench 1-100Vals AI28.5%†14.8%†
Vibe Code Bench v1.1Vals AI88.4%†76.2%†
SWE Atlas: Codebase QnAScale AI SEAL63.2%†–
SWE Atlas: Test WritingScale AI SEAL62.2%†–
DeepSWE v1.1Epoch AI Benchmarking Hub73.7%67.5%
CursorBenchCursor46.6%36.1%
ALE-BenchEpoch AI Benchmarking Hub2164.6†1508.1†
FrontierSWEEpoch AI Benchmarking Hub52.0%25.3%†
Terminal-Bench 4.0 (AA)Artificial Analysis49.0%13.1%
Terminal-Bench 2.1 (AA)Artificial Analysis89.1%84.3%
Startrise FrontendStartrise82.3%†72.1%†
Agents & tools
Terminal-BenchTerminal-Bench51.8%20.3%†
OSWorld-Verified 2.0OSWorld31.4%–
APEX-AgentsMercor65.8%65.3%†
Arena AgentArena (LMArena)8.20.6†
LiveBench Long AgenticLiveBench62.1%†40.7%†
LiveBench BrowserLiveBench84.0%†64.0%†
LiveBench Agentic CodingLiveBench65.2%57.0%†
Terminal-Bench 2.1 (Vals)Vals AI84.6%†78.3%†
MCP AtlasScale AI SEAL85.8%†–
HiL-BenchScale AI SEAL57.0%†–
τ³-bench Bankingτ²-bench48.7%–
τ³-Banking (AA)Artificial Analysis42.1%44.3%
Analyst Agent (AA)Artificial Analysis53.8%41.3%†
CUA-benchVals AI9.0%–
CyberBenchVals AI65.4%†67.7%†
Finance Agent v2Vals AI58.6%†53.7%†
Harvey's Legal Agent Benchmark v1Vals AI6.7%†15.8%†
Legal Research BenchVals AI55.3%†48.1%†
SkillsBenchVals AI60.4%†55.8%†
SREBenchVals AI12.2%†–
Tax Agent BenchVals AI45.5%†33.1%†
Terminal-Bench 4.0 (Vals)Vals AI53.5%†17.2%†
Terminal-Bench ScienceVals AI27.1%†11.4%†
Time Horizon Index: KSPVals AI18.8%†5.8%†
Arena SearchArena (LMArena)1214†–
BALROGBALROG63.4%–
PostTrainBenchEpoch AI Benchmarking Hub35.0%†–
GBAEvalEpoch AI Benchmarking Hub79.6%†–
Vending-Bench 2Andon Labs11181.9†9047.0†
AutomationBenchArtificial Analysis56.6%63.1%
GDP.pdf (AA)Artificial Analysis21.6%17.8%
EnterpriseOps-GymArtificial Analysis47.5%48.3%†
AA-Briefcase v1.1Artificial Analysis16571531†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis28.6%6.2%†
DeepSec-Bench (AA)Artificial Analysis25.9%–
FACTS SearchGoogle85.3%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub85.6%66.0%†
FrontierMath Tier 4Epoch AI Benchmarking Hub73.2%31.7%†
OTIS Mock AIMEEpoch AI Benchmarking Hub98.9%97.8%†
ProofBenchVals AI99.0%51.0%†
LiveBench MathematicsLiveBench95.7%92.6%†
Arena MathsArena (LMArena)15171445†
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub59.9%49.3%†
LiveBench 2.0 Data AnalysisLiveBench74.6%†75.3%†
LiveBench ResearchLiveBench88.9%†84.5%†
LiveBench 1.0 Data AnalysisLiveBench74.5%73.9%†
AA-OmniscienceArtificial Analysis37.128
MMLU-ProVals AI91.6%†89.4%†
LegalBenchVals AI87.0%†86.3%†
CorpFinVals AI73.2%†66.2%†
TaxEvalVals AI75.1%†71.1%†
BioMysteryBenchVals AI79.3%†72.2%†
Excel Modeling BenchmarkVals AI73.6%†62.7%†
MedCodeVals AI63.6%†44.7%†
MedScribeVals AI91.0%†86.5%†
Public Benefits Bench v1.1Vals AI76.9%†66.8%†
DrugDiscoveryBenchScale AI SEAL53.7%–
EBR-benchEpoch AI Benchmarking Hub45.7%30.5%†
GDP.pdfEpoch AI Benchmarking Hub24.0%16.0%†
AA-Omniscience: accuracyArtificial Analysis60.9%41.9%
AA-Omniscience: non-hallucinationArtificial Analysis39.2%76.0%
Instruction following
LiveBench LanguageLiveBench88.7%83.7%†
LiveBench Instruction FollowingLiveBench63.8%71.9%†
Arena Instruction FollowingArena (LMArena)14951452†
Human preference
Arena TextArena (LMArena)14911456†
EQ-Bench 4EQ-Bench1360†–
Arena Creative WritingArena (LMArena)14741443†
Arena Multi-turnArena (LMArena)14871448†
Arena Longer QueriesArena (LMArena)15031475†
Multimodal
Arena VisionArena (LMArena)1320†1263†
LiveBench DesignLiveBench83.0%†79.8%†
MMMU-ProArtificial Analysis84.7%–
MMMU-Pro (Vals)Vals AI89.9%†–
MortgageTaxVals AI72.1%†64.2%†
SAGEVals AI49.4%†28.9%†
Blueprint-Bench 2Epoch AI Benchmarking Hub30.4%†33.2%†
Long context
AA-LCRArtificial Analysis79.3%81.0%
Arena DocumentArena (LMArena)1490†1461†
MLCRArtificial Analysis11.1%18.3%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub163.9†156.6†
LiveBench 2.0LiveBench82.1%†74.9%†
LiveBench 1.0LiveBench80.1%78.0%†
AA Intelligence Index v4.3.2Artificial Analysis50.842.8
Vals RSI IndexVals AI33.0%25.5%†
Vals Science IndexVals AI47.9%†38.1%†
Vals Multimodal IndexVals AI73.9%†–
Safety & honesty
FORTRESSScale AI SEAL16.8%†–

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

What changed

  • Claude Opus 5 (max): 1 new result (Vals RSI Index).
  • Claude Opus 5 (max): 1 score revision.
  • Grok 4.6 (medium): 1 new result (MLCR).
  • Claude Opus 5 (max): 1 new result (ForecastBench); 1 score revision.
  • Claude Opus 5 (max): BenchLeader Index 68.3 → 68.9.
  • Claude Opus 5 (max): 4 new results (GDPval-AA v2.1, AA-Omniscience: accuracy, AA-Omniscience: non-hallucination…).
  • Grok 4.6 (medium): 15 new results (AA Intelligence Index v4.3.2, GPQA Diamond (AA), Humanity's Last Exam (AA)…).
  • Claude Opus 5 (max): 1 score revision.
  • Claude Opus 5 (max): 2 score revisions.
  • Grok 4.6 (medium): 1 score revision.

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

Claude Opus 5 vs Grok 4.6: questions

Is Claude Opus 5 better than Grok 4.6?
Claude Opus 5 (max) and Grok 4.6 (medium) are neck and neck on quality (68.8 vs 64.2, within the ±5.1 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-25, so check the category scores for your use.
Which is cheaper, Claude Opus 5 or Grok 4.6, and will that change?
Grok 4.6 is cheaper today: $3.00 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 Grok 4.6 for coding?
Claude Opus 5 scores higher in coding (70 vs 62 on the category index, where 50 is average). On SciCode, the top shared coding benchmark, it is 56.4% for Claude Opus 5 against 55.9% for Grok 4.6.
Is Claude Opus 5 better than Grok 4.6 for agents and tool use?
Grok 4.6 scores higher in agentic tasks (74 vs 64 on the category index, where 50 is average). On τ³-Banking (AA), the top shared agents & tools benchmark, it is 42.1% for Claude Opus 5 against 44.3% for Grok 4.6.
Which is better for long documents?
They are level on long-context benchmarks (65 vs 66). Claude Opus 5 accepts more context: 1,000,000 against 500,000 tokens for Grok 4.6.
Which has lower latency?
Claude Opus 5 (max) answers first: 24.1 s against 25.5 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 Grok 4.6?
Grok 4.6 streams faster: 78 against 57 output tokens per second.
Do independent benchmarks agree on Claude Opus 5 vs Grok 4.6?
Across the 18 benchmarks both have, Claude Opus 5 leads 12, Grok 4.6 leads 6. Every overall index that rates both puts Claude Opus 5 ahead.
When was Grok 4.6 released?
Grok 4.6 was released on 11 Aug 2026.