Inference
EN

GPT-5.5 vs Kimi K3.1 Pro

Change models
GPT-5.5OpenAI
Kimi K3.1 ProMoonshot AISmartest open
IndexAll categories, out of 100.
69.1 ±2.5
75.7 ±4.7 best
GapIndex points below the top.
−6.6
Top
CoverageIndex weight that has results.
95% best
80%
SpeedOutput tokens per second.
52/s best
43/s
EloLMArena rating and rank.
1466
—
ReleasedMonth of release.
2026-04
2026-10
Input / 1MUS dollars per 1M input tokens.
$5.00
$2.00 best
Output / 1MUS dollars per 1M output tokens.
$30.00
$10.00 best
Batch output / 1MBatch price per 1M output tokens.
$15.00
$6.00 best
1K chats2K tokens in, 500 out, each.
$25.00
$9.00 best
1K documents20K tokens in, 1K out, each.
$130.00
$50.00 best
1K agent tasks200K tokens in, 20K out, each.
$1,600.00
$600.00 best
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
65.8
76.7 best
CodingCode writing and repair.
65.5
71.9 best
ReasoningLogic problems and puzzles.
74.1
77.7 best
MultimodalTasks with images and text.
66.8
72.0 best
KnowledgeFacts and expert knowledge.
70.4
71.0 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
64.8
—
MathMath problems.
74.7
—

Benchmarks

20 shared
Only GPT-5.5: 92Only Kimi K3.1 Pro: 43
GPT-5.5
Kimi K3.1 Pro
Artificial Analysis Humanity's Last ExamKnowledge
45.8%26 / 222
52.5% best11 / 222
Artificial Analysis Long Context ReasoningReasoning
84.3%7 / 217
86.3% best3 / 217
Critical Physics TasksReasoning
27.1% best17 / 214
24.6%22 / 214
Artificial Analysis Omniscience AccuracyKnowledge
58.0% best10 / 164
48.9%23 / 164
Graduate-Level Google-Proof Q&AKnowledge
93.6%8 / 141
94.2% best5 / 141
GDPval-AA normalizedAgentic
42.5%44 / 133
67.3% best3 / 133
Artificial Analysis SciCodeCoding
55.8%27 / 114
60.8% best9 / 114
Artificial Analysis MMMU-ProMultimodal
79.9%23 / 113
83.9% best9 / 113
GPQA Diamond · BenchLMKnowledge
93.6%8 / 84
94.2% best6 / 84
Humanity's Last ExamKnowledge
52.2%19 / 77
70.0% best2 / 77
BrowseCompAgentic
84.4%19 / 63
92.1% best3 / 63
Humanity's Last Exam without toolsKnowledge
41.4%26 / 53
51.1% best8 / 53
Massive Multi-discipline Multimodal Understanding ProMultimodal
81.2%10 / 53
84.1% best2 / 53
MazeBench: visual spatial reasoning in a 3D open worldAgentic
0.0%23 / 48
2.0% best15 / 48
FrontierCode 2.0Coding
22.6% best18 / 27
19.6%22 / 27
Artificial Analysis AnalystAgentAgentic
50.0% best10 / 26
38.8%23 / 26
OSWorld 2.0Agentic
13.0%21 / 25
79.6% best1 / 25
ExploitGymAgentic
13.4%18 / 23
20.7% best13 / 23
OfficeQA ProMultimodal
54.1%13 / 16
67.4% best4 / 16
MMMU-Pro with PythonMultimodal
83.2%
90.0%