Inference
EN

Gemini 3.8 Flash vs Kimi K3.1 Pro

Change models
Gemini 3.8 FlashGoogle
Kimi K3.1 ProMoonshot AISmartest open
IndexAll categories, out of 100.
67.8 ±2.5
75.7 ±4.7 best
GapIndex points below the top.
−7.9
Top
CoverageIndex weight that has results.
95% best
80%
SpeedOutput tokens per second.
68/s best
43/s
EloLMArena rating and rank.
1500
—
ReleasedMonth of release.
2026-09
2026-10
Input / 1MUS dollars per 1M input tokens.
$0.825 best
$2.00
Output / 1MUS dollars per 1M output tokens.
$4.13 best
$10.00
Batch output / 1MBatch price per 1M output tokens.
$1.88 best
$6.00
1K chats2K tokens in, 500 out, each.
$3.71 best
$9.00
1K documents20K tokens in, 1K out, each.
$20.63 best
$50.00
1K agent tasks200K tokens in, 20K out, each.
$247.50 best
$600.00
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
65.3
76.7 best
CodingCode writing and repair.
66.2
71.9 best
ReasoningLogic problems and puzzles.
68.3
77.7 best
MultimodalTasks with images and text.
67.1
72.0 best
KnowledgeFacts and expert knowledge.
70.9
71.0 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
91.3
—
MathMath problems.
64.3
—

Benchmarks

22 shared
Only Gemini 3.8 Flash: 84Only Kimi K3.1 Pro: 41
Gemini 3.8 Flash
Kimi K3.1 Pro
Artificial Analysis Humanity's Last ExamKnowledge
47.8%20 / 222
52.5% best11 / 222
Artificial Analysis Long Context ReasoningReasoning
81.3%34 / 217
86.3% best3 / 217
Critical Physics TasksReasoning
18.3%34 / 214
24.6% best22 / 214
Artificial Analysis Omniscience AccuracyKnowledge
50.5% best19 / 164
48.9%23 / 164
GDPval-AA normalizedAgentic
46.8%34 / 133
67.3% best3 / 133
Artificial Analysis SciCodeCoding
56.6%23 / 114
60.8% best9 / 114
Artificial Analysis MMMU-ProMultimodal
85.6% best4 / 113
83.9%9 / 113
DeepSWEAgentic
73.8%12 / 59
75.5% best5 / 59
MazeBench: visual spatial reasoning in a 3D open worldAgentic
4.0% best14 / 48
2.0%15 / 48
Artificial Analysis AnalystAgentAgentic
46.3% best15 / 26
38.8%23 / 26
Bug Hunt BenchCoding
18.0%23 / 25
36.0% best11 / 25
OSWorld 2.0Agentic
59.0%9 / 25
79.6% best1 / 25
Artificial Analysis AutomationBenchAgentic
59.9%13 / 23
73.3% best4 / 23
Artificial Analysis GDP.pdfAgentic
21.0%15 / 22
31.4% best2 / 22
Artificial Analysis Terminal-Bench v4.0Agentic
19.7%18 / 22
67.2% best1 / 22
Medical Long Context Reasoning (MLCR-AA)Reasoning
10.6%22 / 22
38.3% best4 / 22
Artificial Analysis DeepsecBench-AAAgentic
14.5%16 / 21
36.7% best3 / 21
Artificial Analysis CWE-Bench-AAAgentic
52.5%13 / 19
60.8% best6 / 19
Artificial Analysis Cyber IndexAgentic
25.1%15 / 19
64.1% best1 / 19
Artificial Analysis CyberGym-E2E-AAAgentic
8.4%14 / 19
94.7% best1 / 19