Claude Opus 4.6

Published

Feb 5, 2026

Claude Opus 4.6 is an AI model developed by Anthropic (United States), first published in February 2026. It works in the language, multimodal and vision domain, on tasks such as code generation, language modeling/generation, quantitative reasoning and 15 more.

Epoch AI has no training-compute estimate for this model.

Access: API access. Its weights are not openly released. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Anthropic
Country of organization: United States
Domain: Language, Multimodal, Vision
Task: Code generation, Language modeling/generation, Quantitative reasoning, Search, Visual question answering, Translation, Image captioning, Instruction interpretation, Mathematical reasoning, Visual puzzles, Code autocompletion, Chat, Character recognition (OCR), Language modeling, Language generation, Text autocompletion, Retrieval-augmented generation, System control
Model accessibility: API access
Open weights: No
Epoch confidence: Likely

Benchmark results

01	WebDev Arena	Arena Score	1555.3
02	ALE Bench	Performance	996.5
03	Epoch Capabilities Index	ECI Score	155.4
04	SWE-bench Verified	mean_score	0.79
05	Terminal-Bench	Accuracy mean	0.7
06	SimpleBench	Score (AVG@5)	0.68
07	WeirdML	Accuracy	0.66
08	SimpleQA Verified	mean_score	0.43
09	FrontierMath	mean_score	0.38
10	GSO	Score OPT@1	0.33
11	Posttrainbench	Average (%)	0.23
12	Humanity's Last Exam	Accuracy	0.19
13	FrontierMath Tier 4	mean_score	0.15
14	Enigma Eval	Accuracy	0.07