Glimmer Index
Plus $4/mo

Claude Opus 4

Anthropic · Released May 22, 2025 · Closed weights

Flagship Claude model for deep reasoning, coding, and long-horizon agents

Strongest at
Writing
Top 38% of the past year's models · one benchmark so far
Context window
200K
tokens in · 32K out
Lab announcement ↗Open in the timeline

Capability profile

Scored against 91 models released in the past year. 50 is the average model; further out is better.

This modelMedian of the past year
5090Median of the past yearCoding: 52 · Better than 37%Writing: 58 · Top 38%Reasoning: 50 · Better than 30%Math: 54 · Better than 43%Science: 50 · Better than 8%CodingWritingReasoningMathScience
Coding52Better than 37%
Writing58Top 38% · 1 test
Reasoning50Better than 30%
Math54Better than 43% · 1 test
Science50Better than 8%

No results yet: Reading & documents, Visual reasoning, Computer use

Where these scores come from

Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.

The Claude Opus line

8 releases in the Claude Opus line.

Claude Opus 4May 22, 202575dClaude Opus 4.1Aug 5, 2025111dClaude Opus 4.5Nov 24, 202573dClaude Opus 4.6Feb 5, 202670dClaude Opus 4.7Apr 16, 202642dClaude Opus 4.8May 28, 202657dClaude Opus 5Jul 24, 2026

A newer release is out: Claude Opus 5.5.

ECI STANDING#105 of 270
SCORE142.52140.29–144.3 CI
FRONTIER LAGLaunched 1.7 months behindScores about what Gemini 2.5 Pro (Mar 2025) did in March 2025.
SCORES LIKE THESE MODELSQwen 3.6 FlashAlibaba · Apr 2026142.89Qwen3.5-35B-A3BAlibaba · Feb 2026142.49GPT-5.5 InstantOpenAI · May 2026142.43Qwen3-MaxAlibaba · Sep 2025142.33
Full Epoch Capabilities Index ranking ↗

Epoch benchmark boards

Best result on each board it has run. The tick marks the leader.

MATH level 5
85%#24 of 98
SimpleBench
59%#30 of 89
Cybench
38%#8 of 22
METR Time Horizons
64%#16 of 42
DTBench
82%#78 of 179
WeirdML
44%#70 of 141
DeepResearch Bench
47%#13 of 26
ARC-AGI-2
9%#56 of 103
6 more boards
Fiction.LiveBench
61%#32 of 58
VPCT
38%#19 of 29
GSO-Bench
7%#23 of 35
ARC-AGI
36%#73 of 107
SWE-Bench verified
71%#24 of 32
GeoBench
49%#24 of 28

Anthropic's past year

12 release days in the past year, about one every 30 days. The next is expected around Nov 10, 2026. Forecasts ↗

NovDecJanFebMarAprMayJunJulAugSepOctNov

Specifications

RELEASE DATEMay 22, 2025
WEIGHTSClosed weights
CONTEXT WINDOW200K
MAX OUTPUT32K
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext, image/jpeg, image/png, image/gif, image/webp, application/pdf
OUTPUTStext

Models.dev record ↗

Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.