Claude Sonnet 5.5

  • Closed weights
  • Reasoning
  • Input: text · image · file
  • Released 2026-09
RankedEnough results for an index.
IndexAll categories, out of 100.75.7±2.5Median 50.7
CoverageIndex weight that has results.95%
SpeedOutput tokens per second.107/sMedian 76/s
First tokenSeconds to the first output token.1.5 sMedian 1.1 s
Input / 1MUS dollars per 1M input tokens.$2.00batch $1.00Median $0.60
Output / 1MUS dollars per 1M output tokens.$10.00batch $5.00 Batch price per 1M output tokens.Median $2.50
ContextMaximum tokens in one request.1MMedian 262K
EloLMArena rating and rank.1470±6 #35Median 1425

Details

From OpenRouter
API IDThe model ID on OpenRouter.
anthropic/claude-sonnet-5.5
Max outputMaximum output tokens per request.
128K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.10
Weights
—
RetiresDate OpenRouter removes the model.
—

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.21 counted75.67 / 221
CodingCode writing and repair.15 counted78.76 / 214
ReasoningLogic problems and puzzles.10 counted76.414 / 225
MultimodalTasks with images and text.6 counted72.76 / 88
KnowledgeFacts and expert knowledge.7 counted72.315 / 266
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted41.8—
MathMath problems.5 counted74.513 / 171

Results

65 counted
BenchmarkResultThe score from the source.PlacePlace among results on this benchmark.
OTIS Mock AIME 2024-2025Math100.0%1 / 203
Code MigrationCoding69.8% ±4.31 / 77
LiveBench CodingCoding91.4%1 / 66
ProofBench v1.1Math100.0%1 / 55
Vibe Code Bench v1.1Coding92.4% ±1.33 / 130
MirrorCodeCoding97.8% ±1.51 / 40
GDPval-AA normalizedAgentic66.9%4 / 132
GPQA diamond · Epoch AIKnowledge95.6% ±1.47 / 220
Critical Physics TasksReasoningAbove fitFar above this model's index.31.4%7 / 207
SWE-bench ProCoding81.3%3 / 83
LiveBench DesignAgentic85.0%2 / 52
Terminal-Bench 4.0Agentic64.1% ±1.03 / 68
HealthBench ProfessionalKnowledge69.2%1 / 21
Artificial Analysis SciCodeCoding61.0%6 / 113
LiveBench Long AgenticAgentic78.8%3 / 51
Agent Arena task outcomeAgentic17.64 / 59
Terminal-Bench 2.1 · Vals AIAgentic83.1% ±1.76 / 75
Terminal-Bench 4.0.0Agentic61.8% ±1.52 / 23
LiveBench ReasoningReasoning91.6%7 / 80
Terminal-Bench 4.1.0Agentic63.9% ±1.55 / 52
Agent Arena steerabilityAgentic14.86 / 59
ARC-AGI-1 (semi-private)Reasoning96.5%14 / 132
Mystery Game PuzzlesReasoning65.0% ±4.89 / 83
LiveBench Coding 2.0Coding94.7%6 / 52
LiveBench MathematicsMath96.1%8 / 66
FrontierMath-Tiers-1-3-v2-PrivateMath88.8% ±1.912 / 93
Toolathlon-VerifiedAgentic77.8%3 / 23
Chess PuzzlesReasoning59.0% ±4.913 / 93
ARC-AGI-2 (semi-private)Reasoning83.9%20 / 137
FrontierMath-Tier-4-v2-PrivateMath80.5% ±6.314 / 75
cursorBench40Coding55.5%4 / 21
LiveBench ResearchAgentic83.6%10 / 52
IOICoding83.1% ±3.79 / 46
FrontierSWE v2Coding61.9%5 / 25
Bug Hunt BenchCoding51.3%6 / 30
LiveBench Browser UseAgentic84.0%11 / 49
Furniture AssemblyReasoning75.0% ±5.59 / 40
OSWorld 2.1 (all 108 tasks)Agentic80.1%3 / 12
FrontierCode 2.0Coding42.2%8 / 32
ARC-AGI-3 (semi-private)Reasoning24.3%9 / 30
DeepSWEAgentic71.0%23 / 64
Terminal-Bench-Science 0.1Agentic59.9%6 / 15
LiveBench Agentic CodingAgentic56.3%27 / 66
SimpleQA VerifiedKnowledge46.5% ±1.642 / 94
FrontierCode 1.1 MainCoding46.2%8 / 17
OfficeQA ProMultimodal65.6%8 / 16
Chartography without toolsMultimodal61.6%6 / 12
Artificial Analysis GDP.pdfAgentic25.8%12 / 23
LiveBench LanguageKnowledge78.0%38 / 66
cursorBench41Coding54.4%7 / 12
CWE-bench v2Agentic61.4%12 / 20
LiveBench Instruction FollowingInstruction56.8%60 / 66
LiveBench Data AnalysisReasoning59.5%73 / 80
MedScribeHealthcare91.1% ±2.03 / 111
EMBFinance75.7% ±2.43 / 75
BioMysteryBenchScience81.1% ±0.62 / 34
ProgramBenchCoding6.5% ±1.74 / 66
MedCodeHealthcare52.9% ±2.112 / 108
Tax Agent BenchFinance42.0% ±3.612 / 89
ARC-AGI-1 (public eval)Reasoning97.8%16 / 116
Terminal-Bench-ScienceScience47.1% ±6.07 / 48
MysteryMechanismScience49.1% ±3.45 / 34
ARC-AGI-2 (public eval)Reasoning90.1%19 / 127
Finance Agent (v2)Finance58.1% ±0.712 / 79
Legal Research BenchLegal48.1% ±3.515 / 98
FrontierMath-ErdosMath2.9% ±2.13 / 16
SRE BenchOther30.2% ±2.88 / 37
Public Benefits Bench v1.1Public benefits67.2% ±1.215 / 50
SAGEEducation88.0% ±1.610 / 30
HLAB-Vals-1.1Legal10.4% ±2.625 / 42
CyberBench v1.1Other59.6% ±5.238 / 49
SpireClimbOther1.3% ±0.910 / 11
Global MMLUMultilingual92.1%—
OfficeQAMultimodal76.9%—
HealthBench raw scoreKnowledge69.4%—
PhysicianBenchKnowledge63.2%—
LatchBio SingleCellBenchKnowledge59.1%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseVals AINo licence statedEpoch AICC BYARC Prize FoundationTerms restrict reuseLiveBenchNo licence statedTerminal-BenchNo licence statedLMArenaCC BY 4.0models.devMITCheaper InferenceNo licence stated

History

13 releases
Index78.4 → 75.7
23 Oct 202621 Nov 2026
Output / 1M$10.00 → $10.00
23 Oct 202621 Nov 2026

Questions

How much does Claude Sonnet 5.5 cost?
$2.00 per 1M input tokens, $10.00 per 1M output tokens.
How good is Claude Sonnet 5.5?
Its index is 75.7 out of 100. That is place 12 of 282 ranked models.
What is Claude Sonnet 5.5 best at?
Coding: 78.7 out of 100, place 6 of 214.
How fast is Claude Sonnet 5.5?
107 output tokens per second, 1.5 s to the first token.
What is the context window of Claude Sonnet 5.5?
1M tokens.
Is Claude Sonnet 5.5 open source?
No. Its weights are not published.
When was Claude Sonnet 5.5 released?
September 2026.

Badge

Claude Sonnet 5.5 on Inference360: place 12 of 282, index 75.7

Compare with

Muse Wonder Full75.8GPT-6.2 Sol76.0GPT-6.1 Sol76.2Gemini 4 Hydrogen76.2GPT-6.1 Terra75.1

Lower price

GPT-6.1 Terra75.1 · $5.00/1MGemini 4 Hydrogen76.2 · $7.50/1M

More from Anthropic

All models
Claude Opus 5.577.1Claude Haiku 5.565.3Claude Fable 5.174.5Claude Mythos 5.1UnrankedClaude Fable 5.284.0