Inference
EN

Inkling 2 Small vs Kimi K3

Change models
Inkling 2 SmallThinking Machines Lab
Kimi K3Moonshot AI
IndexAll categories, out of 100.
66.4 ±3.8
66.9 ±2.4 best
GapIndex points below the top.
−0.5 · no clear gap
Top
CoverageIndex weight that has results.
90%
95% best
SpeedOutput tokens per second.
—
44/s
EloLMArena rating and rank.
1403
1474 best
ReleasedMonth of release.
2026-11
2026-07
Input / 1MUS dollars per 1M input tokens.
$0.30
$0.192 best
Output / 1MUS dollars per 1M output tokens.
$1.00 best
$16.00
Batch output / 1MBatch price per 1M output tokens.
—
$11.40
1K chats2K tokens in, 500 out, each.
$1.10 best
$8.38
1K documents20K tokens in, 1K out, each.
$7.00 best
$19.84
1K agent tasks200K tokens in, 20K out, each.
$80.00 best
$358.40
ContextMaximum tokens in one request.
1M
1.05M best
AgenticMulti-step tasks with tools.
67.6 best
65.9
CodingCode writing and repair.
66.6 best
66.2
ReasoningLogic problems and puzzles.
68.4 best
67.7
MultimodalTasks with images and text.
63.6
65.6 best
KnowledgeFacts and expert knowledge.
61.5
66.3 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
71.0
MathMath problems.
60.4
66.6 best

Benchmarks

43 shared
Only Inkling 2 Small: 15Only Kimi K3: 86
Inkling 2 Small
Kimi K3
Artificial Analysis Humanity's Last ExamKnowledge
44.1%29 / 222
46.9% best22 / 222
Artificial Analysis Long Context ReasoningReasoning
83.3%13 / 217
88.7% best1 / 217
Artificial Analysis Omniscience AccuracyKnowledge
33.8%71 / 164
47.6% best28 / 164
Graduate-Level Google-Proof Q&AKnowledge
93.0%11 / 141
93.5% best10 / 141
ARC-AGI-2 (semi-private)Reasoning
81.7% best20 / 131
60.4%40 / 131
ARC-AGI-1 (semi-private)Reasoning
96.5% best11 / 126
94.5%21 / 126
Vibe Code Bench v1.1Coding
81.5%28 / 123
85.0% best17 / 123
ARC-AGI-2 (public eval)Reasoning
84.2% best20 / 122
61.7%37 / 122
ARC-AGI-1 (public eval)Reasoning
97.4% best20 / 116
94.9%34 / 116
Artificial Analysis SciCodeCoding
52.3%41 / 114
59.5% best11 / 114
Artificial Analysis MMMU-ProMultimodal
81.6% best15 / 113
80.5%18 / 113
Legal Research BenchLegal
46.6% best18 / 90
46.2%19 / 90
Terminal-Bench 2.1 · BenchLMAgentic
85.1%19 / 86
88.3% best9 / 86
GPQA Diamond · BenchLMKnowledge
93.0%12 / 84
93.5% best10 / 84
Tax Agent BenchFinance
38.9% best11 / 82
33.4%19 / 82
Humanity's Last ExamKnowledge
53.2%16 / 77
56.0% best13 / 77
ProgramBenchCoding
0.5%19 / 65
2.0% best11 / 65
BrowseCompAgentic
90.2%9 / 63
91.2% best7 / 63
Terminal-Bench 4.0Agentic
12.6%41 / 61
17.2% best37 / 61
DeepSWEAgentic
72.1% best17 / 59
67.5%28 / 59