Claude Code
Anthropic's agent, the category's reference point. Full-screen TUI since v2.1.89 - the alt-screen switch broke tmux scrollback and copy-mode for many users (open issues 42670, 67289, 62890). Measured 23.1M tokens on our suite at 8/10 resolved, the most expensive harness we have run.
| GitHub stars | 148K (updated September 2026) |
|---|---|
| Language | TypeScript |
| License | Proprietary |
| Bring your own key | No |
| Pricing | Anthropic subscription (Pro/Max) or API billing; Claude models only |
| Footprint | npm package, Node.js runtime |
| Sandbox | Approval prompts per action |
| Tokens on our 10-task suite | 23.1M (resolved 8/10) |
Repository: anthropics/claude-code
On our benchmark suite
10-task SWE-bench Verified subset, five agents through the same LiteLLM proxy on Z.ai GLM-5.3, same provider, 600 s per-task cap, vanilla configs. Claude Code measured 23.1M tokens, resolving 8/10 of tasks. Methodology and per-task data: full report.