anthropic

Claude Opus 4.7

claude-opus

Stronger Opus tier for advanced software work and high-stakes reasoning

ReasoningTool CallingAttachments

Context Window

1M

Max Output

128K

Temperature

No

Open Weights

No

Knowledge Cutoff

2026-01-31

Released

2026-04-16

Last Updated

2026-04-16

Modalities

Input:TextImagePDF
→
Output:Text

Benchmarks

BenchmarkScoreMetricHarness
SWE-Bench Pro64.3resolve rate-
Terminal-Bench66.1success rateTerminus-2
SWE-Atlas Refactoring48.57scoreClaude Code
Artificial Analysis Coding Agent Index66.6average pass@1Claude Code
SWE-Atlas Codebase QnA81pass@1Claude Code
SWE-Bench Pro44.9pass@1Claude Code
Terminal-Bench73.8pass@1Claude Code
Artificial Analysis Coding Agent Index61.2average pass@1Cursor CLI
SWE-Atlas Codebase QnA78.4pass@1Cursor CLI
SWE-Bench Pro34.4pass@1Cursor CLI
Terminal-Bench70.6pass@1Cursor CLI
Artificial Analysis Coding Agent Index59.9average pass@1Claude Code
SWE-Atlas Codebase QnA71.7pass@1Claude Code
SWE-Bench Pro36.4pass@1Claude Code
Terminal-Bench71.4pass@1Claude Code

Available Providers (5)

ProviderInput /1MOutput /1MCache Read /1MCache Write /1MReasoningStatus
AnyAPIAnyAPI————toggle, effortExperimental
Merge GatewayMerge GatewayRp 89.482Rp 447.408Rp 8.949Rp 111.852toggle, effortExperimental
NEAR AI CloudNEAR AI CloudRp 89.482Rp 447.408Rp 8.949Rp 111.852toggle—
Cloudflare AI GatewayCloudflare AI GatewayRp 89.482Rp 447.408Rp 8.949Rp 111.852effort—
Perplexity AgentPerplexity AgentRp 89.482Rp 447.408Rp 8.949—effort—
← Kembali ke Katalog