[BUG] Native installer binary crashes with SIGILL (Illegal instruction) on non-AVX2 CPUs — request non-AVX2 build
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
The native installer downloads and completes setup, but the binary crashes immediately on launch with an Illegal instruction (SIGILL) error. Root cause: my CPU does not support the AVX2 instruction set, and the native binary appears to be AVX2-compiled. Verified via cat /proc/cpuinfo | grep avx2, which returns empty output, confirming no AVX2 support.
What Should Happen?
The installer should either detect CPU capabilities and install a compatible binary, or at minimum provide a non-AVX2 / baseline x86-64 build variant so the native version can run on older or low-spec CPUs. Currently only the npm-based install works as a fallback. I am running older version of claude for now.
Error Messages/Logs
Setting up Claude Code...
bash: line 183: 26319 Illegal instruction "$binary_path" install ${TARGET:+"$TARGET"}
Installation was killed before it could finish (exit code 132).
Steps to Reproduce
- On a Linux machine with a non-AVX2 CPU (verify with: cat /proc/cpuinfo | grep avx2 — should return empty)
- Run: curl -fsSL https://claude.ai/install.sh | bash
- Script downloads and completes setup
- Binary crashes immediately with "Illegal instruction", exit code 132
Claude Model
None
Is this a regression?
No, this never worked
Last Working Version
Binary Never worked . Last working version 2.1.112
Claude Code Version
2.1.112 ( last working version )
Platform
Anthropic API
Operating System
Ubuntu/Debian Linux
Terminal/Shell
Xterm
Additional Information
CPU does not support AVX2 (confirmed via /proc/cpuinfo). Requesting a non-AVX2 baseline build so the native installer works on older/budget hardware, similar to how other tools (e.g. Node.js) ship multiple CPU-feature build variants.
This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗