Claude Code

license

Proprietary(closed-source CLI/app, requires Claude subscription or Console account)

model

Claude(proprietary, API-only)

note

Agent SDK published but the product is closed SaaS

source

closed(anthropics/claude-code ships plugins, examples and scripts

Closed proprietary product gated behind an Anthropic subscription; the coding agent itself is not open source. A direct read of the repo shows it: the GitHub contents API lists only .claude-plugin, .claude, .devcontainer, Script, examples, plugins, scripts and docs at the root, and LICENSE.md is a bare all-rights-reserved notice pointing at Anthropic's Commercial Terms. The published Agent SDK is a client that spawns the closed CLI, and a client SDK over a closed service counts as closed rather than partly open, the same treatment other hosted products on this map get. Whether a core has been carved out is only a meaningful question where the source is public, so nothing is recorded on it here.

• https://code.claude.com/docs/en/overview recorded 2026-06-04

  requires Claude subscription or Anthropic Console account; proprietary surfaces

• https://api.github.com/repos/anthropics/claude-code/contents/ recorded 2026-08-11

  Root listing of anthropics/claude-code: .claude-plugin, .claude, .devcontainer, .gitattributes, .github, .gitignore, .vscode, CHANGELOG.md, LICENSE.md, README.md, SECURITY.md, Script, demo.gif, examples, feed.xml, plugins, scripts. No source tree for the CLI agent.

• https://raw.githubusercontent.com/anthropics/claude-code/main/LICENSE.md recorded 2026-08-11

  Entire file: "(c) Anthropic PBC. All rights reserved. Use is subject to Anthropic's Commercial Terms of Service." No grant to the source.

Reuters reported Claude Code above roughly $2.5B run-rate revenue, with business subscriptions quadrupled over the year and enterprise now more than half of Claude Code revenue. No clean user count is published, so the band rests on reported commercial traction rather than a counted figure; revenue at that scale implies well into the 1-10M+ developer band.

• https://code.claude.com/docs/en/overview recorded 2026-08-13

  broad multi-surface paid product across terminal, VS Code, desktop app, web, Slack and mobile; most surfaces require a Claude subscription or Anthropic Console account

• https://www.saastr.com/anthropic-just-hit-14-billion-in-arr-up-from-1-billion-just-14-months-ago/ recorded 2026-08-13

  Anthropic $14B ARR; section headed "Claude Code - $2.5 Billion ARR. Doubled Since January."

Frontier-defining coding agent: it pairs the leading SWE-bench Verified models with a mature multi-surface harness (sub-agents, hooks, MCP, scheduling), and is one of the one or two capability frontier-definers in this category. The placement rests on the Marktechpost benchmark-driven ranking, which puts Claude Code first of the field at 87.6% SWE-bench Verified on Opus 4.7 and 80.8% on Opus 4.6. Two of the three cited sources no longer serve the figure to a fetcher - llm-stats.com answers a bot-check interstitial rather than the leaderboard, and the Anthropic Opus 4.8 announcement renders its comparison table outside the fetched HTML - so that ranking is what carries the score.

• https://llm-stats.com/benchmarks/swe-bench-verified recorded 2026-06-04

  Claude Opus 4.8 88.6% / 4.7 87.6% lead SWE-bench Verified. Not re-fetchable - the host now answers a bot-check interstitial instead of the leaderboard.

• https://www.anthropic.com/news/claude-opus-4-8 recorded 2026-08-13

  Anthropic primary announcement of Claude Opus 4.8, the headline coding model behind Claude Code, stating improvements across benchmarks and a dynamic-workflows feature in Claude Code. The benchmark table itself is not in the fetched HTML.

• https://www.marktechpost.com/2026/05/15/best-ai-agents-for-software-development-ranked-a-benchmark-driven-look-at-the-current-field/ recorded 2026-08-13

  ranks Claude Code first of the field - SWE-bench Verified 87.6% on Opus 4.7 and 80.8% on Opus 4.6, Terminal-Bench 2.0 69.4%, and describes it as the leader on code-quality metrics