Inference
EN

Claude Haiku 5.5 vs Claude Sonnet 5.5

Change models
Claude Haiku 5.5Anthropic
Claude Sonnet 5.5Anthropic
IndexAll categories, out of 100.
66.3 ±2.8
77.2 ±2.6 best
GapIndex points below the top.
−10.9
Top
CoverageIndex weight that has results.
95%
95%
SpeedOutput tokens per second.
159/s best
107/s
EloLMArena rating and rank.
1399
1469 best
ReleasedMonth of release.
2026-10
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.10 best
$2.00
Output / 1MUS dollars per 1M output tokens.
$0.50 best
$10.00
Batch output / 1MBatch price per 1M output tokens.
$0.25 best
$5.00
1K chats2K tokens in, 500 out, each.
$0.45 best
$9.00
1K documents20K tokens in, 1K out, each.
$2.50 best
$50.00
1K agent tasks200K tokens in, 20K out, each.
$30.00 best
$600.00
ContextMaximum tokens in one request.
1M
1M
AgenticMulti-step tasks with tools.
68.6
77.5 best
CodingCode writing and repair.
71.0
80.8 best
ReasoningLogic problems and puzzles.
67.8
77.8 best
MultimodalTasks with images and text.
51.9
74.4 best
KnowledgeFacts and expert knowledge.
54.5
71.3 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
37.2
41.5 best
MathMath problems.
68.3
76.1 best

Benchmarks

78 shared
Only Claude Haiku 5.5: 6Only Claude Sonnet 5.5: 50
Claude Haiku 5.5
Claude Sonnet 5.5
Artificial Analysis Humanity's Last ExamKnowledge
44.4%28 / 222
55.0% best6 / 222
Artificial Analysis Long Context ReasoningReasoning
82.7%23 / 217
82.7%23 / 217
GPQA diamond · Epoch AIKnowledge
89.6%44 / 217
95.6% best4 / 217
Critical Physics TasksReasoning
18.9%33 / 214
31.4% best7 / 214
OTIS Mock AIME 2024-2025Math
98.9%22 / 200
100.0% best1 / 200
Artificial Analysis Omniscience AccuracyKnowledge
34.7%68 / 164
51.5% best17 / 164
GDPval-AA normalizedAgentic
56.0%15 / 133
66.9% best4 / 133
ARC-AGI-2 (semi-private)Reasoning
61.3%38 / 131
83.9% best17 / 131
ARC-AGI-1 (semi-private)Reasoning
89.0%39 / 126
96.5% best11 / 126
Vibe Code Bench v1.1Coding
90.4%4 / 123
92.4% best2 / 123
ARC-AGI-2 (public eval)Reasoning
64.4%31 / 122
90.1% best16 / 122
ARC-AGI-1 (public eval)Reasoning
90.1%45 / 116
97.8% best14 / 116
Artificial Analysis SciCodeCoding
55.0%29 / 114
61.0% best6 / 114
MedScribeHealthcare
89.5%6 / 109
91.1% best3 / 109
MedCodeHealthcare
51.5%15 / 106
52.9% best12 / 106
SimpleQA VerifiedKnowledge
23.8%75 / 91
46.5% best39 / 91
Chess PuzzlesReasoning
55.0%16 / 90
59.0% best10 / 90
FrontierMath-Tiers-1-3-v2-PrivateMath
75.1%23 / 90
88.8% best10 / 90
Legal Research BenchLegal
43.8%25 / 90
48.1% best12 / 90
Tax Agent BenchFinance
31.9%27 / 82
42.0% best10 / 82