BenchLeader

Grok 4.7 vs Kimi K3

Full profiles: Grok 4.7 (SpaceXAI) and Kimi K3 (Moonshot AI): every benchmark, price and rank.

Updated · Data from 16 sources

As of 16 Nov 2026, Grok 4.7 and Kimi K3 are neck and neck on quality (57.4 vs 60.7 on the BenchLeader Index, inside the ±6.5 margin), and Grok 4.7 is 2.0× cheaper at $3.00 vs $6.00 per 1M tokens blended. Choose Grok 4.7 for agents & tools (57 vs 54); choose Kimi K3 for fast first answers (0.8 s vs 55.9 s) and maths (72 vs 56), knowledge (66 vs 60), and human preference (69 vs 63).

Key facts (as of 16 Nov 2026)

  • Grok 4.7 scores 57.4 on the BenchLeader Index and Kimi K3 scores 60.7 (BenchLeader, 16 Nov 2026).
  • Grok 4.7 costs $3.00 per 1M tokens blended (list price, 16 Nov 2026).
  • Kimi K3 costs $6.00 per 1M tokens blended (list price, 16 Nov 2026).
  • Of the 11 benchmarks both models have, Kimi K3 leads 9 (BenchLeader, 16 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
CodingEither61 vs 62 category score; SciCode 57.4% vs 58.7%Either
Agents & tool useGrok 4.757 vs 54 category score; Terminal-Bench 2.1 (Vals) 73.4% vs 80.9%Medium
MathsKimi K372 vs 56 category score; LiveBench Mathematics 84.4% vs 95.7%High
KnowledgeKimi K366 vs 60 category score; LiveBench 1.0 Data Analysis 78.7% vs 76.9%Medium
Chat quality (human preference)Kimi K369 vs 63 on Arena votesMedium
Lowest costGrok 4.7$3.00 vs $6.00 per 1M blended; agentic coding session $0.144 vs $0.240–
Fastest first answerKimi K30.8 s (Kimi K3) vs 55.9 s (Grok 4.7 (xhigh)) 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

  • Grok 4.7 (xhigh) and Kimi K3 are neck and neck on quality (57.4 vs 60.7, within the ±6.5 margin).
  • Grok 4.7 (xhigh) is stronger in agents & tools.
  • Kimi K3 is stronger in coding, human preference, knowledge, maths.
  • Grok 4.7 (xhigh) is 2.0× cheaper ($3.00 vs $6.00 per 1M blended).

At a glance

SpecGrok 4.7Kimi K3
DeveloperSpaceXAIMoonshot AI
Released21 Sep 202616 Jul 2026
Context window (tokens)500,0001,048,576
Compared atGrok 4.7 (xhigh)Kimi K3
Open weightsNoYes
Price in / out / cached input per 1M$2.00 / $6.00 / $0.500$3.00 / $15.00 / $0.300

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.

CategoryGrok 4.7 (xhigh)Kimi K3Leader
Reasoning55.6–not comparable
Coding60.961.8level
Agents & tools56.953.5Grok 4.7
Maths55.871.5Kimi K3
Knowledge59.966.1Kimi K3
Human preference62.568.6Kimi K3
Multimodal55.5–not comparable
Long context62.9–not comparable
Composite52.8–not 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 avg75CodingAgents & toolsMathsKnowledgeHumanpreferenceGrok 4.7, Coding: 60.9Grok 4.7, Agents & tools: 56.9Grok 4.7, Maths: 55.8Grok 4.7, Knowledge: 59.9Grok 4.7, Human preference: 62.5Kimi K3, Coding: 61.8Kimi K3, Agents & tools: 53.5Kimi K3, Maths: 71.5Kimi K3, Knowledge: 66.1Kimi K3, Human preference: 68.6

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

Model, click to add or removeIndex

Where sources disagree

Across the 11 benchmarks both have, Grok 4.7 leads 2, Kimi K3 leads 9. Independent sources split: Grok 4.7 leads on AA Intelligence Index v5.0.1 and Vals Index; Kimi K3 leads on Arena Text.

Overall indexGrok 4.7 (xhigh)Kimi K3Leader
BenchLeader Index57.460.7neck and neck
AA Intelligence Index v5.0.135.517.4†Grok 4.7
Vals Index58.653.0†Grok 4.7
Arena Text14431487†Kimi K3

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.

MeasureGrok 4.7 (xhigh)Kimi K3
Blended price per 1M tokens$3.00$6.00
Input / output per 1M$2.00 / $6.00$3.00 / $15.00
Cached input per 1M$0.500$0.300
Price per task: chat reply$0.0026$0.0057
Price per task: summarize a 30-page report$0.028$0.045
Price per task: code edit$0.021$0.041
Price per task: agentic coding session$0.144$0.240
Price per task: structured extraction$0.0052$0.0090
Agentic coding session, input mostly cached$0.076$0.118
Cost per benchmark task (AA Index)$7.77–
Output tokens per task (AA Index)105k–
Of which reasoning69k–
Time per task (AA Index)26 min–
Output speed68 tok/s76 tok/s
Time to first answer55.9 s0.8 s
Time to a complete answer, last measured 13 Nov 2026145.5 s–

Each model is compared at its best-ranked reasoning-effort setting. Grok 4.7 (xhigh) scores 57.4 and takes 56 s to its first answer token; Grok 4.7 (high) scores 56.9 in 35 s, Grok 4.7 (low) scores 55.4 in 10 s. Kimi K3 scores 60.7 and takes 1 s to its first answer token; Kimi K3 (max) scores 60.2 in 55 s, Kimi K3 (low) scores 53.6 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

183 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.

MetricPublisherGrok 4.7 (xhigh)Kimi K3
BenchLeader IndexBenchLeader57.460.7
Agents & tools scoreBenchLeader56.953.5
Coding scoreBenchLeader60.961.8
Composite scoreBenchLeader52.8–
Human preference scoreBenchLeader62.568.6
Knowledge scoreBenchLeader59.966.1
Long context scoreBenchLeader62.9–
Maths scoreBenchLeader55.871.5
Multimodal scoreBenchLeader55.5–
Reasoning scoreBenchLeader55.6–
Blended price $/MBenchLeader$3.00$6.00
Output speedBenchLeader68 tok/s76 tok/s
Time to first answerBenchLeader55.9 s0.8 s
Context windowBenchLeader500,000 tokens1,048,576 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub92.7%93.1%†
SimpleBenchSimpleBench–60.7%†
Arena Hard PromptsArena (LMArena)14661516†
LiveBench 2.0 ReasoningLiveBench82.7%90.7%†
LiveBench 1.0 ReasoningLiveBench82.7%90.7%
GPQA Diamond (AA)Artificial Analysis–93.5%†
Humanity's Last Exam (AA)Artificial Analysis43.1%46.9%†
HLE-Diamond (AA)Artificial Analysis23.2%26.9%†
GPQA Diamond (Vals)Vals AI–92.9%
ARC-AGI-1ARC Prize89.5%94.5%†
ARC-AGI-2ARC Prize61.4%60.4%†
ARC-AGI-3ARC Prize7.0%–
CritPtArtificial Analysis17.7%23.4%†
CritPt v2 hard subset (AA)Artificial Analysis33.5%40.5%†
MysteryMechanismVals AI25.2%–
HLE-Diamond (Scale)Scale AI SEAL23.4%†22.2%
Arena ExpertArena (LMArena)14781528†
Chess PuzzlesEpoch AI Benchmarking Hub41.0%36.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub29.0%26.0%†
LMCAEpoch AI Benchmarking Hub49.4%52.7%†
DTBenchEpoch AI Benchmarking Hub96.0%91.2%†
ForecastBenchForecastBench–64.2%†
Coding
SciCodeSciCode57.4%58.7%
WeirdMLWeirdML–82.6%†
FrontierCodeCognition47.6%†44.2%
Show all 183 benchmarks
MetricPublisherGrok 4.7 (xhigh)Kimi K3
Coding
Arena CodingArena (LMArena)14931538†
Arena WebDevArena (LMArena)16411658†
LiveBench 2.0 CodingLiveBench73.1%53.1%†
LiveBench 1.0 CodingLiveBench77.2%81.5%
SciCode (AA)Artificial Analysis57.4%59.5%†
LiveCodeBenchVals AI–87.2%†
IOIVals AI57.7%48.9%†
SWE-bench (Vals)Vals AI–93.4%
Code MigrationVals AI44.8%16.1%†
Code Migration v1.1Vals AI53.2%28.6%†
ProgramBenchVals AI0.5%2.0%
SWE-bench MultimodalVals AI42.9%44.2%†
Vibe Code Bench 1-100Vals AI–18.2%†
Vibe Code Bench v1.1Vals AI86.2%85.0%†
Vibe Code Bench 1-100 v1.1Vals AI26.4%–
SWE Atlas v1.1Scale AI SEAL62.1%†61.0%
SWE Atlas v1.1: Codebase QnAScale AI SEAL73.4%†65.3%
SWE Atlas v1.1: Test WritingScale AI SEAL64.4%†63.3%
SWE Atlas v1.1: RefactoringScale AI SEAL48.6%†54.3%
SWE-Bench Pro V2Scale AI SEAL–97.7%†
SWE-Bench Pro V2 HARDScale AI SEAL–88.2%†
MirrorCodeEpoch AI Benchmarking Hub54.4%54.4%†
Surface Evolver BenchEpoch AI Benchmarking Hub–95.0%
DeepSWE v1.1Epoch AI Benchmarking Hub–68.5%†
CursorBenchCursor46.3%–
ALE-BenchEpoch AI Benchmarking Hub1545.81524.5†
FrontierSWEEpoch AI Benchmarking Hub29.5%25.9%†
WeirdML v3WeirdML8.4%7.5%†
Terminal-Bench 4.0 (AA)Artificial Analysis25.8%12.6%†
Terminal-Bench 2.1 (AA)Artificial Analysis–85.0%†
Startrise FrontendStartrise75.9%†79.8%
GameBench 2 OverallDuelLab GameBench 263.6†48.0
GameBench 2 by reasoning settingDuelLab GameBench 271.245.7†
Design Arena WebsiteDesign Arena1244†1336
FrontierCode 1.1 (AA)Artificial Analysis33.8%43.6%†
FrontierCode 2.0 (AA)Artificial Analysis28.7%10.9%†
Agents & tools
Terminal-BenchTerminal-Bench38.8%23.0%†
APEX-AgentsMercor54.6%†47.8%
Arena AgentArena (LMArena)2.60.6†
LiveBench Long AgenticLiveBench45.3%36.0%†
LiveBench BrowserLiveBench49.3%75.2%†
LiveBench Agentic CodingLiveBench54.0%62.2%
Terminal-Bench 2.1 (Vals)Vals AI73.4%80.9%
MCP AtlasScale AI SEAL–82.3%†
HiL-BenchScale AI SEAL–42.7%
τ³-bench Bankingτ²-bench–37.1%†
GDPval-AA v2.1, normalised scoreArtificial Analysis60.7%–
GDPval-AA v2.1Artificial Analysis17171542†
τ³-Banking (AA)Artificial Analysis–46.0%†
ITBench SRE (AA)Artificial Analysis42.1%47.7%†
Analyst Agent (AA)Artificial Analysis52.5%38.8%†
APEX-Agents (AA)Artificial Analysis–41.3%†
CyberBenchVals AI69.5%75.2%†
Finance Agent v2Vals AI52.3%53.1%†
Finance Agent v2.1Vals AI60.6%60.0%†
Harvey's Legal Agent Benchmark v1.1Vals AI18.3%12.9%†
Harvey's Legal Agent Benchmark v1Vals AI12.5%12.9%†
Legal Research BenchVals AI47.1%46.1%†
Robo-benchVals AI0.2%–
SREBenchVals AI3.8%–
Tax Agent BenchVals AI35.8%33.4%†
Terminal-Bench 4.0 (Vals)Vals AI28.8%17.2%†
Terminal-Bench ScienceVals AI10.0%1.4%†
Time Horizon Index: KSPVals AI–10.5%†
Arena Agent Task CompletionArena (LMArena)5.13.4†
PostTrainBenchEpoch AI Benchmarking Hub–31.5%
GBAEvalEpoch AI Benchmarking Hub–48.4%
Vending-Bench 2Andon Labs10536.8†5165.0
AutomationBenchArtificial Analysis65.6%58.3%†
GDP.pdf (AA)Artificial Analysis20.0%22.0%†
Harvey LAB-AA v1.1Harvey9.4%5.3%†
EnterpriseOps-GymArtificial Analysis–45.3%†
AA-Briefcase v1.1Artificial Analysis16361497†
GDPxGDPx by SemiAnalysis3.9%3.7%†
Design Arena AgenticDesign Arena1277†1261
Terminal-Bench-Science 0.1 (AA)Artificial Analysis11.4%8.1%†
CWE-Bench (AA)Artificial Analysis68.3%53.3%†
DeepSec-Bench (AA)Artificial Analysis26.9%12.4%†
CyberGym end-to-end (AA)Artificial Analysis74.0%58.0%†
MKA (AA)Artificial Analysis40.1%–
FACTS SearchGoogle85.6%†90.3%
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub53.0%72.2%†
FrontierMath Tier 4Epoch AI Benchmarking Hub17.1%39.0%†
OTIS Mock AIMEEpoch AI Benchmarking Hub98.1%97.2%†
ProofBenchVals AI26.0%†87.0%
LiveBench MathematicsLiveBench95.7%84.4%
AIME 2026MathArena–96.7%†
HMMT February 2026MathArena–97.0%†
MathArena ApexMathArena–65.6%†
Arena MathsArena (LMArena)14471502†
FrontierMath ErdősEpoch AI Benchmarking Hub0.0%0.0%†
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub56.0%50.6%†
LiveBench 2.0 Data AnalysisLiveBench76.9%78.7%†
LiveBench ResearchLiveBench89.5%79.8%†
LiveBench 1.0 Data AnalysisLiveBench76.9%78.7%
AA-Omniscience v1.1Artificial Analysis31.2-1.7†
MMLU-ProVals AI–88.0%†
LegalBenchVals AI84.4%86.2%†
CorpFinVals AI–71.6%
TaxEvalVals AI–75.7%
PRBench LegalScale AI SEAL47.6%–
BioMysteryBenchVals AI69.3%72.6%†
Excel Modeling BenchmarkVals AI67.0%66.7%†
Excel Modeling Benchmark v1.1Vals AI71.5%71.3%†
MedCodeVals AI49.5%49.4%†
MedScribeVals AI89.4%88.0%†
Public Benefits Bench v1.1Vals AI65.6%68.2%†
Access to Justice BenchmarkVals AI73.7%68.2%†
Financial Coaching BenchVals AI74.1%74.5%†
Immigration BenchVals AI68.0%68.3%†
GDP.pdfEpoch AI Benchmarking Hub22.8%19.0%†
AA-Omniscience v1.1: accuracyArtificial Analysis42.7%42.4%†
AA-Omniscience v1.1: non-hallucinationArtificial Analysis80.0%23.4%†
HealthBench Professional (AA)OpenAI55.8%53.2%†
FACTS ParametricGoogle54.3%†–
Instruction following
LiveBench LanguageLiveBench80.1%85.5%
LiveBench Instruction FollowingLiveBench75.3%71.4%
Arena Instruction FollowingArena (LMArena)14351489†
Human preference
Arena TextArena (LMArena)14431487†
EQ-Bench 4EQ-Bench–1319
EQ-Bench Creative Writing v3EQ-Bench2012†2083
Arena Creative WritingArena (LMArena)14211459†
Arena Multi-turnArena (LMArena)14351502†
Arena Longer QueriesArena (LMArena)14541500†
Multimodal
Arena VisionArena (LMArena)1230–
LiveBench DesignLiveBench75.0%72.5%†
MMMU-ProArtificial Analysis78.2%†80.5%†
MMMU-Pro (Vals)Vals AI–88.2%
MortgageTaxVals AI–66.3%
SAGEVals AI88.7%86.8%†
Humanity's Sixth SenseScale AI SEAL–25.5%†
GeoBenchGeoBench61.0%†–
Blueprint-Bench 2Epoch AI Benchmarking Hub32.5%†29.5%
Furniture AssemblyEpoch AI Benchmarking Hub20.8%34.2%†
VIALS-AAHandshake AI37.9%–
Code2Video-AAArtificial Analysis11271041†
AA-3DArtificial Analysis992984†
FACTS MultimodalGoogle40.9%†–
Long context
AA-LCR v1.1Artificial Analysis76.7%88.7%†
MLCRArtificial Analysis20.0%21.7%†
FACTS GroundingGoogle68.6%†–
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub156.2†158.6
LiveBench 2.0LiveBench67.9%65.0%†
LiveBench 1.0LiveBench77.4%79.2%
AA Intelligence Index v5.0.1Artificial Analysis35.517.4†
AA Intelligence Index v4.3.2Artificial Analysis46.543.6†
AA Progress Index v1Artificial Analysis322.7319.9†
Vals IndexVals AI58.653.0†
Vals Finance IndexVals AI58.8%58.0%†
Vals RSI IndexVals AI20.2%20.9%†
Vals Science IndexVals AI34.8%–
Vals Multimodal IndexVals AI–73.4%
Vals Router IndexVals AI58.6%†53.0%
Vals Legal IndexVals AI72.7%71.7%†
AA Cyber IndexArtificial Analysis56.441.3†
Safety & honesty
FORTRESSScale AI SEAL–27.3%
DistressBenchScale AI SEAL52.4%†73.5%
Metadata
AA Openness IndexArtificial Analysis–38.9†

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • Grok 4.7 (xhigh): BenchLeader Index 58.0 → 57.4.
  • Kimi K3: BenchLeader Index 59.1 → 60.7.
  • Grok 4.7 (xhigh): 8 new results (LiveBench 2.0, LiveBench Long Agentic, LiveBench 2.0 Reasoning…).
  • Kimi K3: 1 score revision.
  • Grok 4.7 (xhigh): 1 new result (MirrorCode).
  • Grok 4.7 (xhigh): 1 new result (ALE-Bench).
  • Grok 4.7 (xhigh): 2 score revisions.
  • Kimi K3: 4 score revisions.
  • Kimi K3: 1 score revision.
  • Kimi K3: 1 new result (FACTS Search).

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

Grok 4.7 vs Kimi K3: questions

Is Grok 4.7 better than Kimi K3?
Grok 4.7 (xhigh) and Kimi K3 are neck and neck on quality (57.4 vs 60.7, within the ±6.5 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
Grok 4.7 vs Kimi K3: which is cheaper, and will that change?
Grok 4.7 is cheaper today: $3.00 against $6.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Grok 4.7 or Kimi K3 better for coding?
They are level in coding (61 vs 62 on the category index). On SciCode, the top shared coding benchmark, it is 57.4% for Grok 4.7 against 58.7% for Kimi K3.
Which is faster, Grok 4.7 or Kimi K3?
Kimi K3 streams faster (76 vs 68 output tokens per second, Artificial Analysis 7-day median).
Is Grok 4.7 or Kimi K3 better for agents and tool use?
Grok 4.7 scores higher in agentic tasks (57 vs 54 on the category index, where 50 is average). On Terminal-Bench 2.1 (Vals), the top shared agents & tools benchmark, it is 73.4% for Grok 4.7 against 80.9% for Kimi K3.
More questions (4)
Which is better for long documents?
Kimi K3 accepts more context: 1,048,576 against 500,000 tokens for Grok 4.7.
Which has lower latency?
Kimi K3 answers first: 0.8 s against 55.9 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 Grok 4.7 vs Kimi K3?
Across the 11 benchmarks both have, Grok 4.7 leads 2, Kimi K3 leads 9. Independent sources split: Grok 4.7 leads on AA Intelligence Index v5.0.1 and Vals Index; Kimi K3 leads on Arena Text.
When was Grok 4.7 released?
Grok 4.7 was released on 21 Sep 2026.