Claude Opus 4.6

ModelActive

by Anthropic · family “claude-opus”

Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire workflows rather than single prompts, making it especially effective... (description from the OpenRouter listing)

in: textin: imagein: fileout: textReasoningTool useStructured output

Epoch Capabilities Index

155.4 #25 of 268

90% CI 153.4 – 157.1

Listing ↗

Released

5 Feb 2026

Context window

1M tokens

Max output

128K tokens

Input $ / 1M tokens

$5.00

Output $ / 1M tokens

$25.0

Cached input $ / 1M

$0.50

Each benchmark shown separately with its own source

Benchmark results (20)

All benchmarks

Benchmark	Domain	Score	vs best recorded	Setting	Run	Source
Furniture Assembly	multimodal	28.3% ±5.9	34%	max	10 Sep 2026	Epoch ↗
LMCA	agents	55.8%	85%	max	—	External ↗
DTBench	reasoning	91.2%	93%	max	—	External ↗
Mystery Game Puzzles	games	25.0% ±4.4	30%	max	25 Jul 2026	Epoch ↗
EBR-bench	reasoning	12.7%	17%	max	29 Jun 2026	Epoch ↗
FrontierMath-Tiers-1-3-v2-Private	math	66.0% ±2.8	70%	max	11 Jun 2026	Eval log ↗
FrontierMath-Tier-4-v2-Private	math	26.8% ±7.0	27%	max	11 Jun 2026	Eval log ↗
CL-bench Life	long-context	17.0%	77%	high	—	External ↗
ProofBench	math	50.0%	50%	max	—	External ↗
APEX-Agents	agents	46.3%	67%	max	—	External ↗
Chess Puzzles	games	14.0% ±3.5	19%	max	6 Aug 2026	Eval log ↗
SimpleQA Verified	knowledge	47.0% ±1.6	62%	max	27 Aug 2026	Eval log ↗
FrontierMath-Tier-4-2025-07-01-Privatesuperseded	math	22.9% ±6.1	48%	max	12 Feb 2026	Epoch ↗
DeepResearch Bench	agents	55.3%	100%	high	—	External ↗
GSO-Bench	coding	41.2%	88%	high	—	External ↗
FrontierMath-2025-02-28-Privatesuperseded	math	40.7% ±2.9	78%	max	12 Feb 2026	Epoch ↗
HLE	knowledge	34.4%	63%	max	—	External ↗
WeirdML	coding	78.0% ±0.0	83%	high	—	External ↗
OTIS Mock AIME 2024-2025	math	91.1% ±4.3	91%	max	6 Aug 2026	Eval log ↗
GPQA diamond	science	88.4% ±2.3	92%	max	6 Aug 2026	Eval log ↗

Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.

API list price over time

Price history

$ per 1M tokens

8 observations (OpenRouter listing + Internet Archive snapshots). History accumulates with every ingest run; a single point means no change has been observed yet.