Claude Opus 4.6

Name: Claude Opus 4.6
Author: Anthropic

Closed

Claude Opus 4.6 is an AI model from Anthropic.

Publisher: Anthropic
Family: Claude
Context: 1M
Input
Output
License: proprietary
Released: 5 Feb 2026

Cite

Notes

Only stored in your browser.

Attribution

Benchmark scores: frontiermathAA LiveBench SWE-benchmatharenaprime-hub OpenReward Vals AI EQ-Bench Lech Mazur

Attribution policy →

Context

Input

/1M tok

Output

$25

/1M tok

Blended

$10

#444/ 461/1M · 3:1

Speed

#115/ 461tok/s

Cache read

$0.50

0.10x input

Cache write

$6.25

1.25x input

Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter

Reasoning effort	Index	Speed
Reasoning, max	44.9	0 t/s
Non-reasoning, high	38.8	0 t/s

Intelligence Index per reasoning-effort setting, via Artificial Analysis · pricing identical across tiers

Reported on 39 evals across 15 domains - best 89.3% on LiveBench - Math · #12 of 58

Reported eval scores

MRCRV293

Codenames2.1

AutomationBench56.1

Infraresolutionbench84.9

KellyBench88771

AIME202599.8

Autonomous Skill Evolution1.5

Skillsbench44.5

Thematic Generalization Benchmark80.6

LiveBench - Reasoning88.7

Terminal-Bench (Hard)48.5

LiveBench - Language83.3

LiveBench - Coding78.2

SWE-bench Multilingual72

LiveBench - Math89.3

LiveBench - Data Analysis69.9

IFBench44.6

LiveBench - Instruction Following63.3

LiveBench78.8

Humanity's Last Exam (HLE)19.1