Kimi K3vsGPT-5.4

Kimi K3
GPT-5.4
Specifications
Parameters
2.8T
Context window
1M
API pricing
Input price
$3.00$2.50
Output price
$15.00$15.00
Cached input price
$0.30$0.25
Cheapest input
$2.125Morph$2.50Azure
Cheapest output
$11.5502Sail Research$15.00Azure
Benchmarks
BullshitBench v2
74%48%
Next.js Evals
84%65%
BrowseComp
91.2%82.7%
Humanity's Last Exam · with tools
56%52.1%
GPQA Diamond
93.5%92.8%
Benchmarks
ProgramBench
0%
DeepSWE 1.1
69%
DeepSWE 1.0
67.5%
Supabase Evals · with skills
78.3%
Supabase Evals · no skills
81.2%
Terminal-Bench 2.1
88.3%
Terminal-Bench 2.0
75.1%
Expert-SWE (Internal)
68.5%
MCP Atlas
84.2%
JobBench
52.9%
Toolathlon-Verified
73.2%
Toolathlon
54.6%
CyberGym
79%
Humanity's Last Exam · no tools
43.5%
ARC-AGI-2
73.95%
FrontierMath · Tier 1–3
47.6%
FrontierMath · Tier 4
27.1%
OSWorld-Verified
75%
GDPval-AA v2
1668
GDPval (win/tie rate)
83%
CharXiv Reasoning
84.8%
MMMU-Pro
81.6%
threejseval
1549
Overview
CompanyMoonshot AIOpenAI
Release dateJul 16 2026Mar 5 2026
AccessOpen WeightProprietary

Other comparisons

Kimi K3vsClaude Fable 5.1GPT-5.4vsClaude Fable 5.1Kimi K3vsGemini 3.8 FlashGPT-5.4vsGemini 3.8 FlashKimi K3vsMuse Spark 1.3GPT-5.4vsMuse Spark 1.3Kimi K3vsGrok 4.6GPT-5.4vsGrok 4.6Kimi K3vsDeepSeek-V4.1-FlashGPT-5.4vsDeepSeek-V4.1-FlashKimi K3vsMistral Medium 3.5GPT-5.4vsMistral Medium 3.5

Frequently asked questions

Kimi K3 leads GPT-5.4 on 5 of the 5 benchmarks they both report. GPT-5.4 is cheaper on input: $2.50 vs $3.00 per million tokens. Both charge $15.00 per million output tokens. GPT-5.4 shipped 133 days before Kimi K3, so benchmark comparisons should account for the intervening progress.

Kimi K3 is open weight, while GPT-5.4 is proprietary.

On BullshitBench v2, Kimi K3 leads at 74% vs GPT-5.4 at 48%. On Next.js Evals, Kimi K3 leads at 84% vs GPT-5.4 at 65%. On BrowseComp, Kimi K3 leads at 91.2% vs GPT-5.4 at 82.7%. On Humanity's Last Exam · with tools, Kimi K3 leads at 56% vs GPT-5.4 at 52.1%. On GPQA Diamond, Kimi K3 leads at 93.5% vs GPT-5.4 at 92.8%.