Inference
EN

GPT-6 Astra vs Kimi K3.1 Pro

Change models
GPT-6 AstraOpenAI
Kimi K3.1 ProMoonshot AISmartest open
IndexAll categories, out of 100.
78.0 ±1.9 best
75.7 ±4.7
GapIndex points below the top.
Top
−2.3 · no clear gap
CoverageIndex weight that has results.
100% best
80%
SpeedOutput tokens per second.
34/s
43/s best
EloLMArena rating and rank.
1440
—
ReleasedMonth of release.
2026-09
2026-10
Input / 1MUS dollars per 1M input tokens.
$10.00
$2.00 best
Output / 1MUS dollars per 1M output tokens.
$50.00
$10.00 best
Batch output / 1MBatch price per 1M output tokens.
$25.00
$6.00 best
1K chats2K tokens in, 500 out, each.
$45.00
$9.00 best
1K documents20K tokens in, 1K out, each.
$250.00
$50.00 best
1K agent tasks200K tokens in, 20K out, each.
$3,000.00
$600.00 best
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
74.6
76.7 best
CodingCode writing and repair.
76.9 best
71.9
ReasoningLogic problems and puzzles.
83.8 best
77.7
MultimodalTasks with images and text.
71.2
72.0 best
KnowledgeFacts and expert knowledge.
78.5 best
71.0
MultilingualTasks in many languages.
73.6
—
InstructionTasks with strict prompt rules.
79.5
—
MathMath problems.
78.0
—

Benchmarks

32 shared
Only GPT-6 Astra: 106Only Kimi K3.1 Pro: 31
GPT-6 Astra
Kimi K3.1 Pro
Artificial Analysis Humanity's Last ExamKnowledge
54.7% best8 / 222
52.5%11 / 222
Artificial Analysis Long Context ReasoningReasoning
80.7%37 / 217
86.3% best3 / 217
Critical Physics TasksReasoning
31.7% best4 / 214
24.6%22 / 214
Artificial Analysis Omniscience AccuracyKnowledge
61.6% best6 / 164
48.9%23 / 164
Graduate-Level Google-Proof Q&AKnowledge
96.0% best1 / 141
94.2%5 / 141
GDPval-AA normalizedAgentic
53.7%25 / 133
67.3% best3 / 133
Artificial Analysis SciCodeCoding
56.5%24 / 114
60.8% best9 / 114
Artificial Analysis MMMU-ProMultimodal
86.9% best2 / 113
83.9%9 / 113
GPQA Diamond · BenchLMKnowledge
96.0% best1 / 84
94.2%6 / 84
BrowseCompAgentic
91.5%6 / 63
92.1% best3 / 63
DeepSWEAgentic
74.1%11 / 59
75.5% best5 / 59
MazeBench: visual spatial reasoning in a 3D open worldAgentic
23.0% best1 / 48
2.0%15 / 48
Humanity's Last Exam with toolsAgentic
57.2%13 / 36
70.0% best1 / 36
FrontierCode 2.0Coding
42.8% best3 / 27
19.6%22 / 27
Artificial Analysis AnalystAgentAgentic
51.2% best8 / 26
38.8%23 / 26
Bug Hunt BenchCoding
45.0% best3 / 25
36.0%11 / 25
OSWorld 2.0Agentic
72.6%2 / 25
79.6% best1 / 25
Artificial Analysis AutomationBenchAgentic
68.5%7 / 23
73.3% best4 / 23
ExploitGymAgentic
42.4% best3 / 23
20.7%13 / 23
Artificial Analysis GDP.pdfAgentic
31.0%3 / 22
31.4% best2 / 22