BenchLeader

Agnes 2.5 Pro Beta vs Kimi K3

Full profiles: Agnes 2.5 Pro Beta (Sapiens AI) and Kimi K3 (Moonshot AI): every benchmark, price and rank.

Updated · Data from 12 sources

As of 16 Nov 2026, there are not enough shared benchmarks for a verdict, and Agnes 2.5 Pro Beta is 40× cheaper at $0.150 vs $6.00 per 1M tokens blended. Agnes 2.5 Pro Beta leads in every category we measure both in.

Key facts (as of 16 Nov 2026)

  • Agnes 2.5 Pro Beta costs $0.150 per 1M tokens blended (list price, 16 Nov 2026).
  • Kimi K3 costs $6.00 per 1M tokens blended (list price, 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
Agents & tool useAgnes 2.5 Pro Beta64 vs 54 category scoreHigh
Chat quality (human preference)–Only Kimi K3 has an Arena rating (69 category score)Not comparable
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • Not enough shared benchmarks for a verdict.
  • Agnes 2.5 Pro Beta is stronger in agents & tools.
  • Agnes 2.5 Pro Beta is 40× cheaper ($0.150 vs $6.00 per 1M blended).

At a glance

SpecAgnes 2.5 Pro BetaKimi K3
DeveloperSapiens AIMoonshot AI
Released26 Aug 202616 Jul 2026
Context window (tokens)1,000,0001,048,576
Compared atAgnes 2.5 Pro BetaKimi K3
Open weightsNoYes
Price in / out / cached input per 1M–$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.

CategoryAgnes 2.5 Pro BetaKimi K3Leader
Coding–61.8not comparable
Agents & tools64.453.5Agnes 2.5 Pro Beta
Maths–71.5not comparable
Knowledge–66.1not comparable
Human preference–68.6not comparable
Multimodal57.7–not comparable
Long context66.2–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 & toolsMathsKnowledgeLong contextMultimodalHumanpreferenceAgnes 2.5 Pro Beta, Agents & tools: 64.4Agnes 2.5 Pro Beta, Long context: 66.2Agnes 2.5 Pro Beta, Multimodal: 57.7Kimi K3, Coding: 61.8Kimi K3, Agents & tools: 53.5Kimi K3, Maths: 71.5Kimi K3, Knowledge: 66.1Kimi K3, Human preference: 68.6

Dashed: that model has no result on the axis it crosses; the line only joins its neighbours.

Not drawn, no result for these models: Reasoning, Instruction following.

Model, click to add or removeIndex

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.

MeasureAgnes 2.5 Pro BetaKimi K3
Blended price per 1M tokens–$6.00
Input / output per 1M– / –$3.00 / $15.00
Cached input per 1M–$0.300
Price per task: chat reply–$0.0057
Price per task: summarize a 30-page report–$0.045
Price per task: code edit–$0.041
Price per task: agentic coding session–$0.240
Price per task: structured extraction–$0.0090
Agentic coding session, input mostly cached–$0.118
Output speed–76 tok/s
Time to first answer–0.8 s

Each model is compared at its best-ranked reasoning-effort setting. Agnes 2.5 Pro Beta. 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

168 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.

MetricPublisherAgnes 2.5 Pro BetaKimi K3
BenchLeader IndexBenchLeader–60.7
Agents & tools scoreBenchLeader64.453.5
Long context scoreBenchLeader66.2–
Multimodal scoreBenchLeader57.7–
Coding scoreBenchLeader–61.8
Human preference scoreBenchLeader–68.6
Knowledge scoreBenchLeader–66.1
Maths scoreBenchLeader–71.5
Blended price $/MBenchLeader–$6.00
Output speedBenchLeader–76 tok/s
Time to first answerBenchLeader–0.8 s
Context windowBenchLeader1,000,000 tokens1,048,576 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub–93.1%†
SimpleBenchSimpleBench–60.7%†
Arena Hard PromptsArena (LMArena)–1516†
LiveBench 2.0 ReasoningLiveBench–90.7%†
LiveBench 1.0 ReasoningLiveBench–90.7%
GPQA Diamond (AA)Artificial Analysis90.5%93.5%†
Humanity's Last Exam (AA)Artificial Analysis37.5%46.9%†
HLE-Diamond (AA)Artificial Analysis–26.9%†
GPQA Diamond (Vals)Vals AI–92.9%
ARC-AGI-1ARC Prize–94.5%†
ARC-AGI-2ARC Prize–60.4%†
CritPtArtificial Analysis15.7%23.4%†
CritPt v2 hard subset (AA)Artificial Analysis–40.5%†
HLE-Diamond (Scale)Scale AI SEAL–22.2%
Arena ExpertArena (LMArena)–1528†
Chess PuzzlesEpoch AI Benchmarking Hub–36.0%†
Mystery Game PuzzlesEpoch AI Benchmarking Hub–26.0%†
LMCAEpoch AI Benchmarking Hub–52.7%†
DTBenchEpoch AI Benchmarking Hub–91.2%†
ForecastBenchForecastBench–64.2%†
Coding
SciCodeSciCode–58.7%
WeirdMLWeirdML–82.6%†
FrontierCodeCognition–44.2%
Arena CodingArena (LMArena)–1538†
Arena WebDevArena (LMArena)–1658†
Show all 168 benchmarks
MetricPublisherAgnes 2.5 Pro BetaKimi K3
Coding
LiveBench 2.0 CodingLiveBench–53.1%†
LiveBench 1.0 CodingLiveBench–81.5%
SciCode (AA)Artificial Analysis48.8%59.5%†
LiveCodeBenchVals AI–87.2%†
IOIVals AI–48.9%†
SWE-bench (Vals)Vals AI–93.4%
Code MigrationVals AI–16.1%†
Code Migration v1.1Vals AI–28.6%†
ProgramBenchVals AI–2.0%
SWE-bench MultimodalVals AI–44.2%†
Vibe Code Bench 1-100Vals AI–18.2%†
Vibe Code Bench v1.1Vals AI–85.0%†
SWE Atlas v1.1Scale AI SEAL–61.0%
SWE Atlas v1.1: Codebase QnAScale AI SEAL–65.3%
SWE Atlas v1.1: Test WritingScale AI SEAL–63.3%
SWE Atlas v1.1: RefactoringScale AI SEAL–54.3%
SWE-Bench Pro V2Scale AI SEAL–97.7%†
SWE-Bench Pro V2 HARDScale AI SEAL–88.2%†
MirrorCodeEpoch AI Benchmarking Hub–54.4%†
Surface Evolver BenchEpoch AI Benchmarking Hub–95.0%
DeepSWE v1.1Epoch AI Benchmarking Hub–68.5%†
ALE-BenchEpoch AI Benchmarking Hub–1524.5†
FrontierSWEEpoch AI Benchmarking Hub–25.9%†
WeirdML v3WeirdML–7.5%†
Terminal-Bench 4.0 (AA)Artificial Analysis–12.6%†
Terminal-Bench 2.1 (AA)Artificial Analysis69.7%85.0%†
Startrise FrontendStartrise–79.8%
GameBench 2 OverallDuelLab GameBench 2–48.0
GameBench 2 by reasoning settingDuelLab GameBench 2–45.7†
Design Arena WebsiteDesign Arena–1336
FrontierCode 1.1 (AA)Artificial Analysis–43.6%†
FrontierCode 2.0 (AA)Artificial Analysis–10.9%†
Agents & tools
Terminal-BenchTerminal-Bench–23.0%†
APEX-AgentsMercor–47.8%
Arena AgentArena (LMArena)–0.6†
LiveBench Long AgenticLiveBench–36.0%†
LiveBench BrowserLiveBench–75.2%†
LiveBench Agentic CodingLiveBench–62.2%
Terminal-Bench 2.1 (Vals)Vals AI–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 Analysis41.5%–
GDPval-AA v2.1Artificial Analysis13351542†
τ³-Banking (AA)Artificial Analysis35.7%46.0%†
ITBench SRE (AA)Artificial Analysis–47.7%†
Analyst Agent (AA)Artificial Analysis–38.8%†
APEX-Agents (AA)Artificial Analysis–41.3%†
CyberBenchVals AI–75.2%†
Finance Agent v2Vals AI–53.1%†
Finance Agent v2.1Vals AI–60.0%†
Harvey's Legal Agent Benchmark v1.1Vals AI–12.9%†
Harvey's Legal Agent Benchmark v1Vals AI–12.9%†
Legal Research BenchVals AI–46.1%†
Tax Agent BenchVals AI–33.4%†
Terminal-Bench 4.0 (Vals)Vals AI–17.2%†
Terminal-Bench ScienceVals AI–1.4%†
Time Horizon Index: KSPVals AI–10.5%†
Arena Agent Task CompletionArena (LMArena)–3.4†
PostTrainBenchEpoch AI Benchmarking Hub–31.5%
GBAEvalEpoch AI Benchmarking Hub–48.4%
Vending-Bench 2Andon Labs–5165.0
AutomationBenchArtificial Analysis–58.3%†
GDP.pdf (AA)Artificial Analysis–22.0%†
Harvey LAB-AA v1.1Harvey–5.3%†
EnterpriseOps-GymArtificial Analysis–45.3%†
AA-Briefcase v1.1Artificial Analysis–1497†
GDPxGDPx by SemiAnalysis–3.7%†
Design Arena AgenticDesign Arena–1261
Terminal-Bench-Science 0.1 (AA)Artificial Analysis–8.1%†
CWE-Bench (AA)Artificial Analysis–53.3%†
DeepSec-Bench (AA)Artificial Analysis–12.4%†
CyberGym end-to-end (AA)Artificial Analysis–58.0%†
FACTS SearchGoogle–90.3%
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub–72.2%†
FrontierMath Tier 4Epoch AI Benchmarking Hub–39.0%†
OTIS Mock AIMEEpoch AI Benchmarking Hub–97.2%†
ProofBenchVals AI–87.0%
LiveBench MathematicsLiveBench–84.4%
AIME 2026MathArena–96.7%†
HMMT February 2026MathArena–97.0%†
MathArena ApexMathArena–65.6%†
Arena MathsArena (LMArena)–1502†
FrontierMath ErdősEpoch AI Benchmarking Hub–0.0%†
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub–50.6%†
LiveBench 2.0 Data AnalysisLiveBench–78.7%†
LiveBench ResearchLiveBench–79.8%†
LiveBench 1.0 Data AnalysisLiveBench–78.7%
AA-Omniscience v1.1Artificial Analysis–-1.7†
MMLU-ProVals AI–88.0%†
LegalBenchVals AI–86.2%†
CorpFinVals AI–71.6%
TaxEvalVals AI–75.7%
BioMysteryBenchVals AI–72.6%†
Excel Modeling BenchmarkVals AI–66.7%†
Excel Modeling Benchmark v1.1Vals AI–71.3%†
MedCodeVals AI–49.4%†
MedScribeVals AI–88.0%†
Public Benefits Bench v1.1Vals AI–68.2%†
Access to Justice BenchmarkVals AI–68.2%†
Financial Coaching BenchVals AI–74.5%†
Immigration BenchVals AI–68.3%†
GDP.pdfEpoch AI Benchmarking Hub–19.0%†
AA-Omniscience v1.1: accuracyArtificial Analysis–42.4%†
AA-Omniscience v1.1: non-hallucinationArtificial Analysis–23.4%†
HealthBench Professional (AA)OpenAI–53.2%†
Instruction following
LiveBench LanguageLiveBench–85.5%
LiveBench Instruction FollowingLiveBench–71.4%
Arena Instruction FollowingArena (LMArena)–1489†
Human preference
Arena TextArena (LMArena)–1487†
EQ-Bench 4EQ-Bench–1319
EQ-Bench Creative Writing v3EQ-Bench–2083
Arena Creative WritingArena (LMArena)–1459†
Arena Multi-turnArena (LMArena)–1502†
Arena Longer QueriesArena (LMArena)–1500†
Multimodal
LiveBench DesignLiveBench–72.5%†
MMMU-ProArtificial Analysis75.4%80.5%†
MMMU-Pro (Vals)Vals AI–88.2%
MortgageTaxVals AI–66.3%
SAGEVals AI–86.8%†
Humanity's Sixth SenseScale AI SEAL–25.5%†
Blueprint-Bench 2Epoch AI Benchmarking Hub–29.5%
Furniture AssemblyEpoch AI Benchmarking Hub–34.2%†
Code2Video-AAArtificial Analysis–1041†
AA-3DArtificial Analysis–984†
Long context
AA-LCR v1.1Artificial Analysis83.0%88.7%†
MLCRArtificial Analysis–21.7%†
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub–158.6
LiveBench 2.0LiveBench–65.0%†
LiveBench 1.0LiveBench–79.2%
AA Intelligence Index v5.0.1Artificial Analysis–17.4†
AA Intelligence Index v4.3.2Artificial Analysis35.243.6†
AA Progress Index v1Artificial Analysis299.1319.9†
Vals IndexVals AI–53.0†
Vals Finance IndexVals AI–58.0%†
Vals RSI IndexVals AI–20.9%†
Vals Multimodal IndexVals AI–73.4%
Vals Router IndexVals AI–53.0%
Vals Legal IndexVals AI–71.7%†
AA Cyber IndexArtificial Analysis–41.3†
Safety & honesty
FORTRESSScale AI SEAL–27.3%
DistressBenchScale AI SEAL–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

  • Kimi K3: BenchLeader Index 59.1 → 60.7.
  • Kimi K3: 1 score revision.
  • Kimi K3: 4 score revisions.
  • Kimi K3: 1 score revision.
  • Kimi K3: 1 new result (FACTS Search).
  • Kimi K3: 1 score revision.
  • Agnes 2.5 Pro Beta: 1 new result (AA Progress Index v1).
  • Kimi K3: 3 score revisions.
  • Kimi K3: 1 score revision.
  • Kimi K3: 1 new result (Vals Router Index).

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

Agnes 2.5 Pro Beta vs Kimi K3: questions

Agnes 2.5 Pro Beta vs Kimi K3: which is cheaper, and will that change?
Agnes 2.5 Pro Beta is cheaper today: $0.150 against $6.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is Agnes 2.5 Pro Beta or Kimi K3 better for agents and tool use?
Agnes 2.5 Pro Beta scores higher in agentic tasks (64 vs 54 on the category index, where 50 is average).
Which is better for long documents?
Both take about 1M tokens: Agnes 2.5 Pro Beta 1,000,000, Kimi K3 1,048,576.
When was Agnes 2.5 Pro Beta released?
Agnes 2.5 Pro Beta was released on 26 Aug 2026.