Claude AI agents build C compiler from scratch

Anthropic researcher Nicholas Carlini published a blog post on Thursday describing how he set 16 instances of the company's Claude Opus 4.6 AI model loose on a shared codebase with minimal supervision. The models were tasked with building a C compiler from scratch. The effort spanned two weeks and nearly 2,000 Claude Code sessions. It cost about 20,000 dollars in API fees. The agents produced a 100,000-line Rust-based compiler capable of building a bootable Linux 6.9 kernel on x86, ARM and RISC-V architectures. Carlini used a new agent teams feature launched with Claude Opus 4.6. Each model instance ran inside its own Docker container, cloned a shared Git repository, claimed tasks by writing lock files and pushed completed code back upstream. No orchestration agent directed the work. Each instance independently identified problems to solve next. The instances resolved merge conflicts on their own. Anthropic released the compiler on GitHub. The compiler can compile major open source projects including PostgreSQL, SQLite, Redis, FFmpeg and QEMU. It achieved a 99 percent pass rate on the GCC torture test suite. It also compiled and ran Doom, which Carlini called the developer's ultimate litmus test. The blog post noted that a C compiler is a near-ideal task for semi-autonomous AI model coding because the specification is decades old and well-defined. Comprehensive test suites already exist, and there is a known good reference compiler to check against.