Skip to content
← Models
ClosedAnthropic/10 trustVerified

Claude Opus 4.7

Anthropic's April 2026 flagship: sharper vision, xhigh reasoning effort, and steadier long-horizon agents.

Visit official page →

About

Claude Opus 4.7 is Anthropic's April 16, 2026 flagship, a direct upgrade to Opus 4.6 aimed at long software engineering runs rather than demo tasks. Anthropic reports 87.6% on SWE-bench Verified (up from 80.8%), 64.3% on SWE-bench Pro, and 69.4% on Terminal-Bench 2.0. Cursor measured it clearing 70% of CursorBench versus 58% for 4.6, and Rakuten saw three times as many production tasks resolved.

The other headline is vision: images up to 2576 pixels on the long edge, about 3x the previous resolution, with XBOW reporting a jump from 54.5% to 98.5% on visual acuity. A new xhigh reasoning effort level and beta task budgets let agentic loops manage their own depth and spend. Pricing held at $5 per million input and $25 per million output tokens. Anthropic is also using it as a test bed for stricter cyber safeguards, including automatic blocking of high-risk cybersecurity requests.

Specs

Context window

1,000,000 tokens (extended context)

Pricing

$5 per million input tokens, $25 per million output tokens

Free quota

API access

Anthropic API (claude-opus-4-7), Claude app and Code, Amazon Bedrock, Google Vertex AI, Microsoft Foundry, GitHub Copilot

Weights

Closed — API only

Released

April 16, 2026

Benchmarks

  • SWE-bench Verified: 87.6%, per Anthropic
  • SWE-bench Pro: 64.3%, per Anthropic
  • Terminal-Bench 2.0: 69.4%, per Anthropic
  • CursorBench: over 70% (per Cursor), vs 58% for Opus 4.6

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

software-engineeringagentic-codingdocument-reasoningvisual-analysis

Trust score — 7/10

/10 trustVerified

Scored 7/10 (verified tier, metrics-v1). Evidence: flagship across Claude surfaces and all major clouds with production use reported by Cursor, Rakuten, and Devin; superseded by Opus 5/5.5 later in 2026. Closed weights. Release facts corroborated across Anthropic announcements and independent outlets; docked for Anthropic's self-reported benchmark figures.

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
8.0/10
Maintenance · recency, releases, issue load
8.0/10
Openness · license + docs
3.0/10
Manual review · sanity read + red-flag check
8.0/10
  • · Adoption 8: flagship across Claude app/Code, Bedrock, Vertex, Foundry, Copilot; Cursor/Rakuten/Devin all reported production use
  • · Maintenance 8: released Apr 16, 2026; superseded by Opus 5/5.5 later in 2026 but remains a documented generation
  • · Openness 3: proprietary, no weights
  • · Manual 8: release facts (Apr 16 2026; claude-opus-4-7; $5/$25; 2576px vision; xhigh effort; SWE-bench Verified 87.6%/Pro 64.3%) corroborated across Anthropic announcements and multiple independent outlets; docked for Anthropic's self-reported benchmark figures

Sources