‹ PublicAI Index
The LLM benchmark aggregator.
Kimi K3
Moonshot · 2.8T · open weights
Strongest in Finance (#7 of 162), weakest in General intelligence (#87 of 178). Above par in 24 of 31 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5 and Claude Fable 5.1 and ahead of Grok 4.6 and DeepSeek V4.1 Flash.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Professional57.2−5.2#13/1942/2
Finance59.2−3.2#7/1621/1
Cybersecurity59.3−4.6#8/471/1
Biology research54.4−16.3#11/311/1
Medical58.6−7.9#17/1491/1
Document design49.1−19.6#19/311/1
Legal56.9−9#28/1731/1
Human preference65−3.8#17/3541/1
Human preference65−3.8#17/3541/1
Reasoning55.4−10#39/1955/5
Science61.8−1.7#13/1221/1
Reasoning56.9−12.1#37/1473/3
Mathematics50.8−13#82/1522/2
Expert reasoning46.8−19.6—/0✱✱
Agents56.4−10#41/2983/6
Long-horizon work49.2−24.8#16/311/1
Web research51−12.9#19/351/2
Computer use56.2−11.9#23/681/2
Knowledge work63.9−6.8#28/2082/2
Workflow automation49.9−16.1—/0✱✱
Core abilities55.4−9.9#42/1992/4
Language59.8−6.8#10/621/1
Data analysis56.8−8.4#13/311/1
Instruction following56.1−17.7#23/621/1
General intelligence49.9−19.5#87/1781/3
Knowledge53.5−9#50/1971/3
Academic knowledge58.4−5.5#22/1231/1
Multimodal understanding35.9−36.2—/19✱0/1
Coding53.2−13#57/1793/6
Agentic coding53.3−13.6#56/1733/5
Repository Q&A46−20.7—/0✱✱
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Kimi K3, left for the other.
§ 3 · Sources
Where the numbers come from
9 publications, 54 figures. Every one links to the page it was read from.
LMArena Text 1487
GDPval-AA 1543
AA-Briefcase 1497
ARC-AGI-2 60.4%
LiveBench 65LiveBench · Agentic 36LiveBench · Reasoning 90.7LiveBench · Coding 53.1LiveBench · Browser Use 75.2LiveBench · Data Analysis 78.7LiveBench · Research 79.8LiveBench · Design 72.5
LiveBench 1.0 79.2LiveBench 1.0 · Reasoning 90.7LiveBench 1.0 · Coding 81.4LiveBench 1.0 · Agentic Coding 62.2LiveBench 1.0 · Mathematics 84.4LiveBench 1.0 · Data Analysis 78.7LiveBench 1.0 · Language 85.5LiveBench 1.0 · Instruction Following 71.4
SimpleBench 60.7%
Vals · Legal Research Bench 46.15%Vals · LegalBench 86.21%Vals · Harvey Legal Agent Benchmark 12.92%Vals · Finance Agent 53.11%Vals · CorpFin 71.56%Vals · TaxEval 75.72%Vals · MortgageTax 66.34%Vals · MedCode 49.36%Vals · MedScribe 88.05%Vals · BioMysteryBench 72.59%Vals · CyberBench 75.24%Vals · SWE-bench Verified 93.4% (Mini-SWE-agent)Vals · Vibe Code Bench 84.97% (OpenHands)Vals · Code Migration 16.1%Vals · GPQA Diamond 92.93%Vals · MMLU Pro 87.97%Vals · ProofBench 87%
GPQA Diamond (Pass@1) 92.9%HLE (Pass@1) 43.5%MathArena Apex (Pass@1) 65.6%Terminal-Bench 2.1 (Pass@1) 88.3%Terminal-Bench 3.0 (Pass@1) 17.7%Terminal-Bench 4.0 (Pass@1) 12.6%DeepSWE v1.1 (Resolved) 67.5%ProgramBench (Almost@1) 17.5%NL2Repo-Bench (Score) 58%CyberGym (Pass@1) 80%HLE w/ tools (Pass@1) 59.8%AutomationBench (Pass@1) 46.7%Agent's Last Exam (Pass@1) 27.6%Chartography w/ tools (Pass@1) 68.1%BabyVision w/ tools (Pass@1) 85.7%ZeroBench-main w/ tools (Pass@5) 41%
Badge
[](https://publicai.io/model-index/m/kimi-k3)