Claude Opus 4.x

weights

closed

data

closed

code

closed

license

Proprietary(API/Bedrock/Vertex/claude.ai)

Anthropic distributes no weights for Opus in any form, serving it only through the API, Bedrock, Vertex and claude.ai. That holds across the whole Opus 4.x line, so it is closed on every dimension regardless of which point release is read.

• https://www.anthropic.com/claude-opus-5-5 recorded 2026-09-29

  Latest Opus release: "Claude Opus 5.5 is now available on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure." Searched the body for weights, self-host, on-prem and license: none.

• https://www.anthropic.com/news/claude-4 recorded 2026-09-29

  Launch post for the Opus 4 line: "Both models are available on our API, Amazon Bedrock, and Google Cloud's Vertex AI." Searched the body for weights, self-host, on-prem and license: none.

• https://www.anthropic.com/news/claude-opus-4-8 recorded 2026-09-29

  Anthropic sells Opus 4.8 as metered API access: "Claude Opus 4.8 is available everywhere today. Pricing for regular usage is unchanged from Opus 4.7: $5 per million input tokens and $25 per million output tokens". Searched the body for weights, open source, license, self-host and on-prem: none.

Opus is Anthropic's premium tier. Its launch post carries enterprise testimonials from partners such as Replit and Vercel but no per-model usage or user figure, so this reading is directional rather than measured.

• https://openrouter.ai/rankings recorded 2026-08-13

  "Live LLM rankings based on benchmarks and real data from millions of people using models through OpenRouter" - the leaderboard itself is client-rendered, so a plain fetch shows the page exists but not which models rank where

• https://www.anthropic.com/news/claude-opus-4-7 recorded 2026-08-13

  named enterprise testimonials (Cursor's Michael Truell on CursorBench, Harvey on BigLaw Bench, Replit's Michele Catasta); no quantitative per-model usage or user figure anywhere on the page

Anthropic reported Opus 4.7 as state of the art on GDPval-AA and Finance Agent, with CursorBench at 70% (up from 58% for Opus 4.6) and BigLaw Bench at 90.9% at high effort. These figures are vendor-reported, and Anthropic has since released Opus 5 and Opus 5.5, which it describes as its leading models.

• https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/ recorded 2026-08-13

  Google's Gemini 3.5 launch post. It names neither Claude nor Opus, so it corroborates nothing about this product.

• https://www.anthropic.com/news/claude-opus-4-7 recorded 2026-08-13

  "On CursorBench, Opus 4.7 is a meaningful jump in capabilities, clearing 70% versus Opus 4.6 at 58%"; "strong substantive accuracy on BigLaw Bench for Harvey, scoring 90.9% at high effort"; "Opus 4.7 is also state-of-the-art on GDPval-AA"