Data◆Article

Grok Build's codebase is 844,530 lines of Rust, comparable in scale to OpenAI Codex's 950,933 lines, revealing that terminal coding agents are far more complex than previously assumed.

Simon Willison measured Grok Build at 844,530 lines of Rust (about 3% vendored) and compares it to Codex's 950,933 lines, concluding terminal coding agents are more complex than he'd realized. ✦ AI generated

Simon Willison · Simon Willison's Weblog · 2026-07-15 · original ↗

Share card ↓Export .md Export .txt

It's quite a surprising codebase! Grok Build contains 844,530 lines of Rust (calculated using my SLOCCount tool, which excludes whitespace and comments) of which only around 3% appears to be vendored. For comparison, openai/codex is 950,933 lines of Rust. It looks like terminal coding agents are significantly more complex than I had realized.

Read full article ↗excerpt · fair-use quotation

✳ Grok Build ✳ openai/codex ✳ Rust

Related moments

Data◆Article · 41 words

Among OpenAI's active internal Codex users, median combined output-token usage grew sharply across departments over six months: Research rose 56x, Customer Support 32x, Engineering 27x, and Legal 13x by June 2026 versus November 2025.

OpenAI Economic Research · Latent Space

Data◆Article · 31 words

Qwen3.6 actually performs better when run through the Codex harness than through its own purpose-built Qwen-Code harness.

Sebastian Raschka · Ahead of AI

Example◆Video · 13:42 · 2m

Jared Sumner rewrote Bun's entire codebase from Zig to Rust by running 64 parallel Claude (Fable pre-release) instances across four work trees, porting and validating files against the existing test suite.

Scott Tolinski · Syntax