Compare · ModelsLive · 2 picked · head to head
Claude Haiku 5.5 vs Kimi K3
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Claude Haiku 5.5 wins on 15/24 benchmarks
Claude Haiku 5.5 wins 15 of 24 shared benchmarks. Leads in agentic · knowledge · general.
Category leads
speed·Kimi K3agentic·Claude Haiku 5.5reasoning·Kimi K3knowledge·Claude Haiku 5.5general·Claude Haiku 5.5coding·Claude Haiku 5.5math·Claude Haiku 5.5
Hype vs Reality
Attention vs performance
Claude Haiku 5.5
#68 by perf·#17 by attention
Kimi K3
#67 by perf·#15 by attention
Best value
Claude Haiku 5.5
27.0x better value than Kimi K3
Claude Haiku 5.5
192.0 pts/$
$0.30/M
Kimi K3
7.1 pts/$
$8.10/M
Vendor risk
Who is behind the model
Anthropic
$965.0B·Tier 1
Moonshot AI
$50.0B·Tier 1
Head to head
24 benchmarks · 2 models
Claude Haiku 5.5Kimi K3
Artificial Analysis · CritPt
Kimi K3 leads by +4.5
Claude Haiku 5.5
18.9
Kimi K3
23.4
Artificial Analysis · GDPval
Claude Haiku 5.5 leads by +4.2
Claude Haiku 5.5
55.9
Kimi K3
51.7
Artificial Analysis · Humanity's Last Exam
Kimi K3 leads by +2.5
Claude Haiku 5.5
44.4
Kimi K3
46.9
Artificial Analysis · Long Context Reasoning
Kimi K3 leads by +6.0
Claude Haiku 5.5
82.7
Kimi K3
88.7
Artificial Analysis · Quality Index
Kimi K3 leads by +0.2
Claude Haiku 5.5
43.4
Kimi K3
43.6
Artificial Analysis · SciCode
Kimi K3 leads by +4.5
Claude Haiku 5.5
55.0
Kimi K3
59.5
APEX-Agents
Claude Haiku 5.5 leads by +8.3
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Claude Haiku 5.5
56.1
Kimi K3
47.8
ARC-AGI
Kimi K3 leads by +5.5
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Claude Haiku 5.5
89.0
Kimi K3
94.5
ARC-AGI-2
Claude Haiku 5.5 leads by +0.8
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Claude Haiku 5.5
61.3
Kimi K3
60.4
Chess Puzzles
Claude Haiku 5.5 leads by +18.9
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Claude Haiku 5.5
52.6
Kimi K3
33.7
Dtbench
Claude Haiku 5.5 leads by +6.3
Claude Haiku 5.5
91.7
Kimi K3
85.3
Frontiercode
Claude Haiku 5.5 leads by +2.2
Claude Haiku 5.5
46.4
Kimi K3
44.2
FrontierMath-Tier-4-v2-Private
Claude Haiku 5.5 leads by +7.3
Claude Haiku 5.5
46.3
Kimi K3
39.0
FrontierMath-Tiers-1-3-v2-Private
Claude Haiku 5.5 leads by +3.1
Claude Haiku 5.5
79.0
Kimi K3
76.0
Frontierswe
Claude Haiku 5.5 leads by +17.9
Claude Haiku 5.5
43.8
Kimi K3
25.9
Furniture Assembly
Claude Haiku 5.5 leads by +19.1
Claude Haiku 5.5
25.0
Kimi K3
6.0
GPQA diamond
Kimi K3 leads by +4.8
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude Haiku 5.5
89.8
Kimi K3
94.6
Lmca
Kimi K3 leads by +2.0
Claude Haiku 5.5
60.1
Kimi K3
62.0
Mirrorcode
Claude Haiku 5.5 leads by +24.5
Claude Haiku 5.5
78.9
Kimi K3
54.4
Mystery Game Puzzles
Claude Haiku 5.5 leads by +4.4
Claude Haiku 5.5
22.9
Kimi K3
18.5
OTIS Mock AIME 2024-2025
Claude Haiku 5.5 leads by +1.7
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude Haiku 5.5
98.9
Kimi K3
97.2
Proofbench
Claude Haiku 5.5 leads by +2.0
Claude Haiku 5.5
89.0
Kimi K3
87.0
SimpleQA Verified
Kimi K3 leads by +26.8
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Claude Haiku 5.5
23.8
Kimi K3
50.6
WeirdML v3
Claude Haiku 5.5 leads by +9.7
Claude Haiku 5.5
26.8
Kimi K3
17.1
Full benchmark table
| Benchmark | Claude Haiku 5.5 | Kimi K3 |
|---|---|---|
Artificial Analysis · CritPt | 18.9 | 23.4 |
Artificial Analysis · GDPval | 55.9 | 51.7 |
Artificial Analysis · Humanity's Last Exam | 44.4 | 46.9 |
Artificial Analysis · Long Context Reasoning | 82.7 | 88.7 |
Artificial Analysis · Quality Index | 43.4 | 43.6 |
Artificial Analysis · SciCode | 55.0 | 59.5 |
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 56.1 | 47.8 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 89.0 | 94.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 61.3 | 60.4 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 52.6 | 33.7 |
Dtbench | 91.7 | 85.3 |
Frontiercode | 46.4 | 44.2 |
FrontierMath-Tier-4-v2-Private | 46.3 | 39.0 |
FrontierMath-Tiers-1-3-v2-Private | 79.0 | 76.0 |
Frontierswe | 43.8 | 25.9 |
Furniture Assembly | 25.0 | 6.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 89.8 | 94.6 |
Lmca | 60.1 | 62.0 |
Mirrorcode | 78.9 | 54.4 |
Mystery Game Puzzles | 22.9 | 18.5 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 98.9 | 97.2 |
Proofbench | 89.0 | 87.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 23.8 | 50.6 |
WeirdML v3 | 26.8 | 17.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.10 | $0.50 | 1.0M tokens (~500 books) | $2.00 | |
| $0.19 | $16.00 | 1.0M tokens (~524 books) | $41.44 |