Inference
EN

DeepSeek V4.1 Flash 1017 vs Kimi K3.1 Pro

Change models
DeepSeek V4.1 Flash 1017DeepSeekFastest
Kimi K3.1 ProMoonshot AISmartest open
IndexAll categories, out of 100.
68.5 ±2.6
75.7 ±4.7 best
GapIndex points below the top.
−7.2
Top
CoverageIndex weight that has results.
95% best
80%
SpeedOutput tokens per second.
261/s best
43/s
EloLMArena rating and rank.
1463
—
ReleasedMonth of release.
2026-10
2026-10
Input / 1MUS dollars per 1M input tokens.
$0.022 best
$2.00
Output / 1MUS dollars per 1M output tokens.
$0.40 best
$10.00
Batch output / 1MBatch price per 1M output tokens.
$0.20 best
$6.00
1K chats2K tokens in, 500 out, each.
$0.244 best
$9.00
1K documents20K tokens in, 1K out, each.
$0.84 best
$50.00
1K agent tasks200K tokens in, 20K out, each.
$12.40 best
$600.00
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
72.2
76.7 best
CodingCode writing and repair.
69.8
71.9 best
ReasoningLogic problems and puzzles.
65.0
77.7 best
MultimodalTasks with images and text.
62.2
72.0 best
KnowledgeFacts and expert knowledge.
64.2
71.0 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
68.3
—
MathMath problems.
65.4
—

Benchmarks

25 shared
Only DeepSeek V4.1 Flash 1017: 69Only Kimi K3.1 Pro: 38
DeepSeek V4.1 Flash 1017
Kimi K3.1 Pro
Artificial Analysis Humanity's Last ExamKnowledge
39.2%51 / 222
52.5% best11 / 222
Artificial Analysis Long Context ReasoningReasoning
84.0%8 / 217
86.3% best3 / 217
Critical Physics TasksReasoning
14.3%52 / 214
24.6% best22 / 214
Artificial Analysis Omniscience AccuracyKnowledge
46.4%33 / 164
48.9% best23 / 164
Graduate-Level Google-Proof Q&AKnowledge
90.9%25 / 141
94.2% best5 / 141
GDPval-AA normalizedAgentic
55.0%21 / 133
67.3% best3 / 133
Artificial Analysis SciCodeCoding
51.9%44 / 114
60.8% best9 / 114
Artificial Analysis MMMU-ProMultimodal
77.0%43 / 113
83.9% best9 / 113
GPQA Diamond · BenchLMKnowledge
90.9%24 / 84
94.2% best6 / 84
Humanity's Last ExamKnowledge
42.6%31 / 77
70.0% best2 / 77
DeepSWEAgentic
77.0% best4 / 59
75.5%5 / 59
MazeBench: visual spatial reasoning in a 3D open worldAgentic
1.0%16 / 48
2.0% best15 / 48
Humanity's Last Exam with toolsAgentic
63.9%5 / 36
70.0% best1 / 36
FrontierCode 2.0Coding
19.3%23 / 27
19.6% best22 / 27
Bug Hunt BenchCoding
39.1% best10 / 25
36.0%11 / 25
ExploitGymAgentic
15.3%16 / 23
20.7% best13 / 23
Agents' Last ExamAgentic
35.4%11 / 21
43.2% best6 / 21
Artificial Analysis DeepsecBench-AAAgentic
31.2%4 / 21
36.7% best3 / 21
Artificial Analysis CWE-Bench-AAAgentic
68.3% best1 / 19
60.8%6 / 19
Artificial Analysis Cyber IndexAgentic
40.8%9 / 19
64.1% best1 / 19