BenchLeader

Kimi K3 vs Step 5 Preview

Full profiles: Kimi K3 (Moonshot AI) and Step 5 Preview (StepFun): every benchmark, price and rank.

Updated · Data from 13 sources

As of 16 Nov 2026, Kimi K3 is ahead of Step 5 Preview on quality (60.7 vs 56.0 on the BenchLeader Index), but Step 5 Preview is 4.2× cheaper at $1.43 vs $6.00 per 1M tokens blended. Kimi K3 leads in every category we measure both in. Step 5 Preview streams ~1.5× faster.

Key facts (as of 16 Nov 2026)

  • Kimi K3 scores 60.7 on the BenchLeader Index and Step 5 Preview scores 56.0 (BenchLeader, 16 Nov 2026).
  • Kimi K3 costs $6.00 per 1M tokens blended (list price, 16 Nov 2026).
  • Step 5 Preview costs $1.43 per 1M tokens blended (list price, 16 Nov 2026).
  • Step 5 Preview streams ~1.5× faster than Kimi K3 (Artificial Analysis, 7-day median, 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
CodingKimi K362 vs 50 category score; SciCode 58.7% vs 58.9%High
Agents & tool useKimi K354 vs 49 category scoreMedium
MathsKimi K372 vs 60 category score; ProofBench 87.0% vs 42.0%High
KnowledgeKimi K366 vs 54 category scoreHigh
Chat quality (human preference)Kimi K369 vs 63 on Arena votesMedium
Lowest costStep 5 Preview$1.43 vs $6.00 per 1M blended; agentic coding session $0.071 vs $0.240–
Fastest first answerKimi K30.8 s (Kimi K3) vs 23.5 s (Step 5 Preview) 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

  • Kimi K3 leads on quality: 60.7 vs 56.0.
  • Kimi K3 is stronger in agents & tools, coding, human preference, knowledge, maths.
  • Step 5 Preview is 4.2× cheaper ($1.43 vs $6.00 per 1M blended).
  • Step 5 Preview streams 1.6× faster (124 vs 76 tokens per second).

At a glance

SpecKimi K3Step 5 Preview
DeveloperMoonshot AIStepFun
Released16 Jul 202618 Sep 2026
Context window (tokens)1,048,5761,000,000
Compared atKimi K3Step 5 Preview
Open weightsYesYes
Price in / out / cached input per 1M$3.00 / $15.00 / $0.300$1.00 / $2.70 / $0.105

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.

CategoryKimi K3Step 5 PreviewLeader
Reasoning–51.7not comparable
Coding61.849.9Kimi K3
Agents & tools53.549.1Kimi K3
Maths71.560.1Kimi K3
Knowledge66.153.7Kimi K3
Human preference68.663.1Kimi K3
Multimodal–59.3not comparable
Long context–69.0not 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 avg75CodingAgents & toolsMathsKnowledgeHumanpreferenceKimi K3, Coding: 61.8Kimi K3, Agents & tools: 53.5Kimi K3, Maths: 71.5Kimi K3, Knowledge: 66.1Kimi K3, Human preference: 68.6Step 5 Preview, Coding: 49.9Step 5 Preview, Agents & tools: 49.1Step 5 Preview, Maths: 60.1Step 5 Preview, Knowledge: 53.7Step 5 Preview, Human preference: 63.1

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 2 benchmarks both have, Kimi K3 leads 1, Step 5 Preview leads 1. Independent sources split: Kimi K3 leads on BenchLeader Index and Arena Text; Step 5 Preview leads on AA Intelligence Index v5.0.1 and Vals Index.

Overall indexKimi K3Step 5 PreviewLeader
BenchLeader Index60.756.0Kimi K3
AA Intelligence Index v5.0.117.4†28.5Step 5 Preview
Vals Index53.0†54.8Step 5 Preview
Arena Text1487†1449Kimi 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.

MeasureKimi K3Step 5 Preview
Blended price per 1M tokens$6.00$1.43
Input / output per 1M$3.00 / $15.00$1.00 / $2.70
Cached input per 1M$0.300$0.105
Price per task: chat reply$0.0057$0.0012
Price per task: summarize a 30-page report$0.045$0.014
Price per task: code edit$0.041$0.010
Price per task: agentic coding session$0.240$0.071
Price per task: structured extraction$0.0090$0.0025
Agentic coding session, input mostly cached$0.118$0.030
Cost per benchmark task (AA Index)–$2.35
Output tokens per task (AA Index)–113k
Of which reasoning–80k
Time per task (AA Index)–15 min
Output speed76 tok/s124 tok/s
Time to first answer0.8 s23.5 s
Time to a complete answer, last measured 13 Nov 2026–54.4 s

Each model is compared at its best-ranked reasoning-effort setting. 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. Step 5 Preview scores 56.0 and takes 24 s to its first answer token. 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

173 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.

MetricPublisherKimi K3Step 5 Preview
BenchLeader IndexBenchLeader60.756.0
Agents & tools scoreBenchLeader53.549.1
Coding scoreBenchLeader61.849.9
Human preference scoreBenchLeader68.663.1
Knowledge scoreBenchLeader66.153.7
Maths scoreBenchLeader71.560.1
Long context scoreBenchLeader–69.0
Multimodal scoreBenchLeader–59.3
Reasoning scoreBenchLeader–51.7
Blended price $/MBenchLeader$6.00$1.43
Output speedBenchLeader76 tok/s124 tok/s
Time to first answerBenchLeader0.8 s23.5 s
Context windowBenchLeader1,048,576 tokens1,000,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub93.1%†–
SimpleBenchSimpleBench60.7%†–
Arena Hard PromptsArena (LMArena)1516†1477
LiveBench 2.0 ReasoningLiveBench90.7%†83.8%†
LiveBench 1.0 ReasoningLiveBench90.7%–
GPQA Diamond (AA)Artificial Analysis93.5%†–
Humanity's Last Exam (AA)Artificial Analysis46.9%†46.5%
HLE-Diamond (AA)Artificial Analysis26.9%†24.3%
GPQA Diamond (Vals)Vals AI92.9%–
ARC-AGI-1ARC Prize94.5%†–
ARC-AGI-2ARC Prize60.4%†–
CritPtArtificial Analysis23.4%†20.9%
CritPt v2 hard subset (AA)Artificial Analysis40.5%†32.5%
MysteryMechanismVals AI–18.5%
HLE-Diamond (Scale)Scale AI SEAL22.2%–
Arena ExpertArena (LMArena)1528†1489
Chess PuzzlesEpoch AI Benchmarking Hub36.0%†–
Mystery Game PuzzlesEpoch AI Benchmarking Hub26.0%†–
LMCAEpoch AI Benchmarking Hub52.7%†–
DTBenchEpoch AI Benchmarking Hub91.2%†–
ForecastBenchForecastBench64.2%†–
Coding
SciCodeSciCode58.7%58.9%
WeirdMLWeirdML82.6%†–
FrontierCodeCognition44.2%–
Arena CodingArena (LMArena)1538†1505
Show all 173 benchmarks
MetricPublisherKimi K3Step 5 Preview
Coding
Arena WebDevArena (LMArena)1658†1573†
LiveBench 2.0 CodingLiveBench53.1%†67.6%†
LiveBench 1.0 CodingLiveBench81.5%–
SciCode (AA)Artificial Analysis59.5%†58.9%
LiveCodeBenchVals AI87.2%†–
IOIVals AI48.9%†38.1%
SWE-bench (Vals)Vals AI93.4%–
Code MigrationVals AI16.1%†39.0%
Code Migration v1.1Vals AI28.6%†47.0%
ProgramBenchVals AI2.0%–
SWE-bench MultimodalVals AI44.2%†40.8%
Vibe Code Bench 1-100Vals AI18.2%†–
Vibe Code Bench v1.1Vals AI85.0%†70.4%
SWE Atlas v1.1Scale AI SEAL61.0%–
SWE Atlas v1.1: Codebase QnAScale AI SEAL65.3%–
SWE Atlas v1.1: Test WritingScale AI SEAL63.3%–
SWE Atlas v1.1: RefactoringScale AI SEAL54.3%–
SWE-Bench Pro V2Scale AI SEAL97.7%†–
SWE-Bench Pro V2 HARDScale AI SEAL88.2%†–
MirrorCodeEpoch AI Benchmarking Hub54.4%†–
Surface Evolver BenchEpoch AI Benchmarking Hub95.0%–
DeepSWE v1.1Epoch AI Benchmarking Hub68.5%†–
ALE-BenchEpoch AI Benchmarking Hub1524.5†–
FrontierSWEEpoch AI Benchmarking Hub25.9%†–
WeirdML v3WeirdML7.5%†–
Terminal-Bench 4.0 (AA)Artificial Analysis12.6%†33.3%
Terminal-Bench 2.1 (AA)Artificial Analysis85.0%†–
Startrise FrontendStartrise79.8%–
GameBench 2 OverallDuelLab GameBench 248.0–
GameBench 2 by reasoning settingDuelLab GameBench 245.7†–
Design Arena WebsiteDesign Arena1336–
FrontierCode 1.1 (AA)Artificial Analysis43.6%†38.4%
FrontierCode 2.0 (AA)Artificial Analysis10.9%†12.0%
Agents & tools
Terminal-BenchTerminal-Bench23.0%†–
APEX-AgentsMercor47.8%–
Arena AgentArena (LMArena)0.6†-3.7
LiveBench Long AgenticLiveBench36.0%†19.4%†
LiveBench BrowserLiveBench75.2%†24.8%†
LiveBench Agentic CodingLiveBench62.2%–
Terminal-Bench 2.1 (Vals)Vals AI80.9%–
MCP AtlasScale AI SEAL82.3%†–
HiL-BenchScale AI SEAL42.7%–
τ³-bench Bankingτ²-bench37.1%†–
GDPval-AA v2.1, normalised scoreArtificial Analysis–54.3%
GDPval-AA v2.1Artificial Analysis1542†1591
τ³-Banking (AA)Artificial Analysis46.0%†–
ITBench SRE (AA)Artificial Analysis47.7%†55.6%
Analyst Agent (AA)Artificial Analysis38.8%†35.0%
APEX-Agents (AA)Artificial Analysis41.3%†38.0%
CyberBenchVals AI75.2%†–
Finance Agent v2Vals AI53.1%†50.7%
Finance Agent v2.1Vals AI60.0%†58.9%
Harvey's Legal Agent Benchmark v1.1Vals AI12.9%†12.9%
Harvey's Legal Agent Benchmark v1Vals AI12.9%†9.2%
Legal Research BenchVals AI46.1%†40.4%
SREBenchVals AI–0.4%
Tax Agent BenchVals AI33.4%†27.3%
Terminal-Bench 4.0 (Vals)Vals AI17.2%†31.8%
Terminal-Bench ScienceVals AI1.4%†2.9%
Time Horizon Index: KSPVals AI10.5%†–
Arena Agent Task CompletionArena (LMArena)3.4†-1.5
PostTrainBenchEpoch AI Benchmarking Hub31.5%–
GBAEvalEpoch AI Benchmarking Hub48.4%–
Vending-Bench 2Andon Labs5165.0–
AutomationBenchArtificial Analysis58.3%†51.0%
GDP.pdf (AA)Artificial Analysis22.0%†14.8%
Harvey LAB-AA v1.1Harvey5.3%†–
EnterpriseOps-GymArtificial Analysis45.3%†47.2%
AA-Briefcase v1.1Artificial Analysis1497†1424
GDPxGDPx by SemiAnalysis3.7%†–
Design Arena AgenticDesign Arena1261–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis8.1%†2.4%
CWE-Bench (AA)Artificial Analysis53.3%†52.5%
DeepSec-Bench (AA)Artificial Analysis12.4%†15.8%
CyberGym end-to-end (AA)Artificial Analysis58.0%†74.8%
MKA (AA)Artificial Analysis–27.6%
FACTS SearchGoogle90.3%–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub72.2%†–
FrontierMath Tier 4Epoch AI Benchmarking Hub39.0%†–
OTIS Mock AIMEEpoch AI Benchmarking Hub97.2%†–
ProofBenchVals AI87.0%42.0%
LiveBench MathematicsLiveBench84.4%–
AIME 2026MathArena96.7%†–
HMMT February 2026MathArena97.0%†–
MathArena ApexMathArena65.6%†–
Arena MathsArena (LMArena)1502†1470
FrontierMath ErdősEpoch AI Benchmarking Hub0.0%†–
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub50.6%†–
LiveBench 2.0 Data AnalysisLiveBench78.7%†63.5%†
LiveBench ResearchLiveBench79.8%†75.5%†
LiveBench 1.0 Data AnalysisLiveBench78.7%–
AA-Omniscience v1.1Artificial Analysis-1.7†16.4
MMLU-ProVals AI88.0%†–
LegalBenchVals AI86.2%†–
CorpFinVals AI71.6%–
TaxEvalVals AI75.7%–
BioMysteryBenchVals AI72.6%†67.4%
Excel Modeling BenchmarkVals AI66.7%†57.1%
Excel Modeling Benchmark v1.1Vals AI71.3%†69.5%
MedCodeVals AI49.4%†–
MedScribeVals AI88.0%†–
Public Benefits Bench v1.1Vals AI68.2%†58.1%
Access to Justice BenchmarkVals AI68.2%†–
Financial Coaching BenchVals AI74.5%†–
Immigration BenchVals AI68.3%†–
GDP.pdfEpoch AI Benchmarking Hub19.0%†–
AA-Omniscience v1.1: accuracyArtificial Analysis42.4%†36.4%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis23.4%†68.7%
HealthBench Professional (AA)OpenAI53.2%†–
Instruction following
LiveBench LanguageLiveBench85.5%–
LiveBench Instruction FollowingLiveBench71.4%–
Arena Instruction FollowingArena (LMArena)1489†1450
Human preference
Arena TextArena (LMArena)1487†1449
EQ-Bench 4EQ-Bench1319–
EQ-Bench Creative Writing v3EQ-Bench2083–
Arena Creative WritingArena (LMArena)1459†1413
Arena Multi-turnArena (LMArena)1502†1445
Arena Longer QueriesArena (LMArena)1500†1461
Multimodal
Arena VisionArena (LMArena)–1263
LiveBench DesignLiveBench72.5%†70.8%†
MMMU-ProArtificial Analysis80.5%†76.4%
MMMU-Pro (Vals)Vals AI88.2%–
MortgageTaxVals AI66.3%–
SAGEVals AI86.8%†–
Humanity's Sixth SenseScale AI SEAL25.5%†–
Blueprint-Bench 2Epoch AI Benchmarking Hub29.5%–
Furniture AssemblyEpoch AI Benchmarking Hub34.2%†–
Code2Video-AAArtificial Analysis1041†821
AA-3DArtificial Analysis984†859
Long context
AA-LCR v1.1Artificial Analysis88.7%†88.3%
MLCRArtificial Analysis21.7%†18.3%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub158.6–
LiveBench 2.0LiveBench65.0%†54.9%†
LiveBench 1.0LiveBench79.2%–
AA Intelligence Index v5.0.1Artificial Analysis17.4†28.5
AA Intelligence Index v4.3.2Artificial Analysis43.6†43.7
AA Progress Index v1Artificial Analysis319.9†318.2
Vals IndexVals AI53.0†54.8
Vals Finance IndexVals AI58.0%†55.4%
Vals RSI IndexVals AI20.9%†–
Vals Science IndexVals AI–29.6%
Vals Multimodal IndexVals AI73.4%–
Vals Router IndexVals AI53.0%–
Vals Legal IndexVals AI71.7%†70.2%
AA Cyber IndexArtificial Analysis41.3†47.7
Safety & honesty
FORTRESSScale AI SEAL27.3%–
DistressBenchScale AI SEAL73.5%–
Metadata
AA Openness IndexArtificial Analysis38.9†33.3

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

  • Kimi K3: BenchLeader Index 59.1 → 60.7.
  • Step 5 Preview: BenchLeader Index 54.2 → 56.0.
  • Step 5 Preview: BenchLeader Index 54.7 → 54.2.
  • Kimi K3: 1 score revision.
  • Step 5 Preview: 1 score revision.
  • Kimi K3: 4 score revisions.
  • Step 5 Preview: 7 score revisions.
  • Kimi K3: 1 score revision.
  • Kimi K3: 1 new result (FACTS Search).
  • Kimi K3: 1 score revision.

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

Kimi K3 vs Step 5 Preview: questions

Is Kimi K3 better than Step 5 Preview?
Kimi K3 leads on quality: 60.7 vs 56.0. The BenchLeader Index combines every independent quality benchmark; Kimi K3 is ahead overall as of 16 Nov 2026, but check the category scores for your use.
Kimi K3 vs Step 5 Preview: which is cheaper, and will that change?
Step 5 Preview is cheaper today: $1.43 against $6.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Kimi K3 or Step 5 Preview better for coding?
Kimi K3 scores higher in coding (62 vs 50 on the category index, where 50 is average). On SciCode, the top shared coding benchmark, it is 58.7% for Kimi K3 against 58.9% for Step 5 Preview.
Which is faster, Kimi K3 or Step 5 Preview?
Step 5 Preview streams ~1.5× faster (124 vs 76 output tokens per second, Artificial Analysis 7-day median).
Is Kimi K3 or Step 5 Preview better for agents and tool use?
Kimi K3 scores higher in agentic tasks (54 vs 49 on the category index, where 50 is average).
More questions (4)
Which is better for long documents?
Both take about 1M tokens: Kimi K3 1,048,576, Step 5 Preview 1,000,000.
Which has lower latency?
Kimi K3 answers first: 0.8 s against 23.5 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.
Do independent benchmarks agree on Kimi K3 vs Step 5 Preview?
Across the 2 benchmarks both have, Kimi K3 leads 1, Step 5 Preview leads 1. Independent sources split: Kimi K3 leads on BenchLeader Index and Arena Text; Step 5 Preview leads on AA Intelligence Index v5.0.1 and Vals Index.
When was Step 5 Preview released?
Step 5 Preview was released on 18 Sep 2026.