head-to-head

OpenAI: GPT-5.6 Luna (batch) vs Anthropic: Claude Sonnet 5 (batch)

Side-by-side comparison of specs, pricing, benchmark scores, and task rankings. Updated 2026-08-13.

	OpenAI: GPT-5.6 Luna (batch)	Anthropic: Claude Sonnet 5 (batch)
Vendor	openai	anthropic
Quality Score	100	100
Benchmark Score	88.2	91.6
Input Price	$0.10/M	$1.00/M
Output Price	$0.60/M	$5.00/M
Context Window	1,050,000	1,000,000
Max Output	128,000	128,000
Tool Calling	✓	✓
Structured Output	✓	✓
Reasoning Mode	✓	✓
Vision	✓	✓
Audio	-	-
Benchmark Scores
ai_index	86.3	91.2
ai_index_agentic	77.4	82.0
ai_index_coding	100.0	100.0

Who wins by task?

Task	OpenAI: GPT-5.6 Luna (batch)	Anthropic: Claude Sonnet 5 (batch)
SQL Generation	172	172
Code Review	167	168
Code Completion	133	120
Code Refactoring	164	165
Bug Fixing	180	182
Unit Test Generation	154	155
Code Documentation	143	143
Regex Writing	137	136
CI/CD Pipelines	144	145
Frontend Component Design	146	147
Data Analysis	171	172
CSV / Spreadsheet Cleanup	154	154
ETL Scripting	154	155
JSON Extraction	149	140
Bulk Data Labeling	135	126
OCR / Document Parsing	146	147
Table Extraction from PDFs	146	147
Long-Document Summarization	160	160
Short-Form Summarization	132	123
Blog Post Writing	140	141

Scores reflect capability match + benchmark data + pricing for each task. Methodology →