Anthropic logo

Claude Opus 4.8

Anthropic · Released May 28, 2026

ProprietarycodingagentsreasoningCompare this model

$5

Input price / MTok

#24 of 27

$25

Output price / MTok

#24 of 27

1.0M

Context window

#6 of 28

65

Tokens / sec

#19 of 21

88.6

SWE-bench Verified

#2 of 12

69.2

SWE-bench Pro

#2 of 9

93.6

GPQA Diamond

#2 of 19

83.4

OSWorld

#2 of 8

Summary

Claude Opus 4.8 is a proprietary language model from Anthropic, released on May 28, 2026.

At $5 input and $25 output per million tokens, it ranks #24 of 27 for input price, placing it among the higher-priced models tracked here.

Its context window of 1.0M tokens ranks #6 of 28 for context depth in this dataset.

It targets coding, agents, reasoning workloads.

Throughput is measured at 65 tokens per second (ranked #19 of 21 for throughput).

Best of the 4.x line for complex agentic and coding work.

Benchmarks

BenchmarkScoreMax%RankSource
SWE-bench Verified88.610088.6%#2/12Anthropic (self-reported in official sys...
SWE-bench Pro69.210069.2%#2/9Anthropic (self-reported in official sys...
GPQA Diamond93.610093.6%#2/19Anthropic (self-reported in official sys...
OSWorld83.410083.4%#2/8Anthropic (self-reported in official sys...

Similar Models

Compare all 6 models

Models with the most shared benchmark coverage and closest scores to Claude Opus 4.8.

Claude Opus 4.8
AnthropicClaude Sonnet 4.6
OpenAIGPT-5.5
AnthropicClaude Fable 5
TencentTencent Hy3
AnthropicClaude Sonnet 5
SWE-bench Verified#2 of 12
Real-world GitHub issue resolution
SWE-bench Pro#2 of 9
Professional software engineering tasks
GPQA Diamond#2 of 19
Graduate-level science reasoning
OSWorld#2 of 8
Computer-use and GUI automation

Compare Claude Opus 4.8 side by side

Pre-filled with 5 similar models

Open comparison

Speed source: https://artificialanalysis.ai/models/claude-opus-4-8