Claude Opus 5.5

Name: Claude Opus 5.5
Author: Anthropic

Multimodal

Anthropic

Claude Opus 5.5 is the first model in Anthropic's Claude 5.5 family. It performs at Claude Fable 5.1 level on most work at about 40% lower cost than Claude Opus 5, and targets long-running agentic coding and knowledge work. It accepts text and image input with text output, supports multilingual and vision workloads, tool use, and adaptive thinking that is always on (API default effort medium). Context window is 1M tokens with 128K max output on the sync Messages API; reliable knowledge and training-data cutoffs are June 2026. First-party pricing is $4/$20 per million input/output tokens, with cache reads at $0.20 and 5-minute cache writes at $5 per million tokens. Fast mode is available in Claude Code and on the Claude Platform at $8/$40 per million input/output tokens with up to 2.5x speed (same model id; not a separate deployment). Capability scores are self-reported from the Claude Opus 5.5 System Card (Table 8.1.A and §8) and the launch announcement. Available on the Claude API as `claude-opus-5-5`.

Key Specifications

Parameters

Context

1.0M

Release Date

September 22, 2026

Average Score

71.1%

API Documentation Research Paper Results Blog

Timeline

Key dates in the model's history

Announcement

September 22, 2026

Last Update

September 23, 2026

Today

October 5, 2026

Technical Specifications

Parameters

Training Tokens

Knowledge Cutoff

June 1, 2026

Family

Capabilities

MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)

$4.00

Output (per 1M tokens)

$20.00

Max Input Tokens

1.0M

Max Output Tokens

128.0K

Supported Features

Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Other Tests

Specialized benchmarks

AA-Briefcase v1.1

System Card Table 8.1.A. AA-Briefcase v1.1 Elo at max effort (max_score 3000). • Self-reported

60.7%

ArXivMath

System Card §8.9. ArXivMath August 2026 (57 problems); without tools; max effort; averaged over four attempts per problem. Evaluated internally without safeguards classifiers. • Self-reported

91.2%

AutomationBench

Opus 5.5 with production safeguards enabled. Zapier AutomationBench; runs without fallback models, so safeguard interventions counted as failures. Adaptive thinking at max effort. • Self-reported

40.0%

BenchCAD

System Card §8.13.2. BenchCAD Vision2Code 1k subset; voxel IoU without tools; adaptive thinking at max effort; averaged over five runs. • Self-reported

73.0%

BenchCAD (with Python tool)

System Card §8.13.2. BenchCAD Vision2Code 1k subset; voxel IoU with tools; adaptive thinking at max effort; averaged over five runs. • Self-reported

96.2%

BioMysteryBench

System Card §8.17.1. BioMysteryBench Human Solvable subset 89.3%; adaptive thinking at max effort; bash/file tools with package allow-list. • Self-reported

89.3%

Chartography

Opus 5.5 with production safeguards enabled. Chartography with tools; adaptive thinking at max effort. • Self-reported

89.0%

CursorBench 4.0

Opus 5.5 with production safeguards enabled. CursorBench 4.0; adaptive thinking at max effort. • Self-reported

57.8%

DeepSWE 1.1

System Card §8.3. DeepSWE v1.1; 5-trial average. Adaptive thinking at max effort. • Self-reported

74.2%

FrontierCode 1.1

Opus 5.5 with production safeguards enabled. FrontierCode v1.1 Main subset; adaptive thinking at max effort. • Self-reported

54.4%

FrontierSWE V2

System Card §8.7. FrontierSWE v2; Proximal harness; max effort; 5 trials per task; mean across trials. • Self-reported

62.3%

GDPval-AA 2.1

Opus 5.5 with production safeguards enabled. GDPval-AA v2.1 Elo (max_score 3000); adaptive thinking at max effort. • Self-reported

61.5%

Global-MMLU

System Card §8.16.1. GMMLU average accuracy 94.3% across 42 languages; adaptive thinking at max effort; single trial; no tools. • Self-reported

94.3%

HealthBench

System Card §8.15.1. HealthBench length-adjusted score 60.6% (penalizes verbose responses); adaptive thinking at max effort; 5-trial average. Prefer length-adjusted over raw 68.1% to match other catalog models and avoid duplicate benchmark_id records. • Self-reported

60.6%

HealthBench Professional

System Card Table 8.1.A length-adjusted score 65.6% (prefer Table 8.1.A over §8.15.2 raw 77.1%). Adaptive thinking at max effort; 5-trial average; safety classifiers enabled with refusal fallback to Claude Opus 5. Length-adjusted per HealthBench Professional paper method. • Self-reported

65.6%

Humanity's Last Exam (no tools, text-only)

System Card Table 8.1.A / §8.11.1. HLE no tools (reasoning-only); adaptive thinking at max effort. Alongside with-tools 67.7% already cataloged from the announcement table. • Self-reported

64.4%

Humanity's Last Exam (with tools, text-only)

Opus 5.5 with production safeguards enabled. Humanity's Last Exam with tools; adaptive thinking at max effort. • Self-reported

67.7%

LatchBio SingleCellBench

System Card §8.17.2. LatchBio SingleCellBench 61.2%; adaptive thinking at max effort; bash/file tools. • Self-reported

61.2%

LatchBio SpatialBench Verified

System Card §8.17.2. LatchBio SpatialBench Verified 72.0%; adaptive thinking at max effort; bash/file tools. • Self-reported

72.0%

Legal Agent Benchmark

System Card §8.14.2. Legal Agent Benchmark held-out set; all-pass rate 8.3% at max effort (AA harness). Mean criterion-pass rate 91.2% noted in card but not stored as a separate catalog metric (all-pass is the standard LAB reporting id). • Self-reported

8.3%

MILU

System Card §8.16.2. MILU average accuracy 93.1% across 11 languages; adaptive thinking at max effort; averaged over 5 trials; no tools. • Self-reported

93.1%

OfficeQA

System Card §8.14.1. OfficeQA full set; max effort; mean of five runs; agentic harness with extracted-text documents and code-execution tools. • Self-reported

78.9%

OfficeQA Pro

System Card §8.14.1. OfficeQA Pro (133-question subset); max effort; mean of five runs. • Self-reported

67.7%

OSWorld 2.0

Opus 5.5 with production safeguards enabled. OSWorld 2.0 partial score (announcement reports partial, not strict); adaptive thinking at max effort. • Self-reported

81.8%

Program Bench

System Card §8.10.1. ProgramBench 166 golden tasks; mini-swe-agent harness; no 6-hour time limit; hidden test pass rate against tests the reference binary passes. • Self-reported

91.2%

SWE-bench Multilingual

System Card Table 8.1.A / §8.2. Adaptive thinking at max effort; 5-trial average. 300 problems across nine programming languages. • Self-reported

93.9%

SWE-Bench Multimodal

System Card Table 8.1.A / §8.2. Adaptive thinking at max effort; 5-trial average. Visual context (screenshots and design mockups). • Self-reported

61.4%

SWE-Bench Pro

System Card Table 8.1.A / §8.2. Adaptive thinking at max effort; 5-trial average. Production safeguards as in card capability summary. • Self-reported

89.9%

Terminal-Bench 4.0

Opus 5.5 with production safeguards enabled. Terminal-Bench 4.0 accuracy 66.4% at xhigh effort (Claude Code harness). Unless otherwise noted, other Opus 5.5 results on this announcement table use adaptive thinking at max effort; TB4 is explicitly reported at xhigh. • Self-reported

66.4%

Terminal-Bench-Science 0.1

Opus 5.5 with production safeguards enabled. Terminal-Bench-Science 0.1; adaptive thinking at max effort. • Self-reported

58.7%

Toolathlon-Verified

System Card §8.14.5. Toolathlon Verified Pass@1 77.8% (3-trial average across 108 tasks). Adaptive thinking at max effort; safety/sandbox stops counted as failures. • Self-reported

77.8%

License & Metadata

License

proprietary

Announcement Date

September 22, 2026

Last Updated

September 23, 2026

Compare Claude Opus 5.5

All comparisons

vs Claude Sonnet 5.5 vs Claude Fable 5.1 vs Claude Opus 5 vs Qwen3.8 Max vs Muse Spark 1.2 vs Muse Spark 1.1 vs Solar Pro 4 vs Atria Dawn Preview

Similar Models

All Models

Claude Sonnet 5.5

Anthropic

Released:Sep 2026

Price:$2.00/1M tokens

Claude Fable 5

Anthropic

Released:Jun 2026

Price:$10.00/1M tokens

Claude Fable 5.1

Anthropic

Released:Sep 2026

Price:$10.00/1M tokens

Claude Sonnet 5

Anthropic

Released:Jun 2026

Price:$2.00/1M tokens

Claude Opus 5

Anthropic

Released:Jul 2026

Price:$5.00/1M tokens

Claude Mythos Preview

Anthropic

Best score:0.9 (GPQA)

Released:Apr 2026

Price:$25.00/1M tokens

Claude Opus 4.6

Anthropic

Best score:1.0 (TAU)

Released:Feb 2026

Price:$5.00/1M tokens

Claude Opus 4.1

Anthropic

Best score:0.8 (TAU)

Released:Aug 2025

Price:$15.00/1M tokens

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.