# GCC 17 freezes features November 15, targets April 2027 stable release

> **Key takeaway:** GCC 17 concludes active feature development on November 15, shifting to bug-fix mode with AVX10, RISC-V native tuning, AMD Zen improvements, and C++29 infrastructure already landed.

**Published:** 2026-11-08T11:12:21+00:00 
**Source:** Phoronix Benchmarks & Drivers 
**Category:** kernel-os 
**Canonical URL:** https://fosswire.org/news/gcc-17-freezes-features-november-15-targets-april-2027-stabl.html 

## Summary
The GNU Compiler Collection will transition from stage one to stage three development on November 15, 2026, halting new feature work for GCC 17 and pivoting to bug fixes and isolated port additions. This follows the project's established annual cadence where November marks the end of active feature development, January enters the stricter stage four for regression fixes only, and April typically brings the stable x.1 release.

GCC 17 introduces several architecture-specific enhancements including AVX10V1AUX support, Function Multi-Versioning for AVX10.2 and APX, 128-bit atomic memory operations, and -mtune=native/-mcpu=native support on RISC-V. The release also adds initial infrastructure for C++29, AMD Zen tuning improvements, SpacemiT X100 core targeting, and preliminary support for AI Compute Extensions.

## What lands in GCC 17 before the freeze
- **AVX10 and x86 extensions:** GCC 17 adds AVX10V1AUX support and Function Multi-Versioning for both AVX10.2 and APX, allowing runtime dispatch of optimized code paths on newer Intel and AMD processors.
- **RISC-V native tuning:** The -mtune=native and -mcpu=native flags now work on RISC-V, enabling automatic optimization for the host CPU without manual architecture specification.
- **128-bit atomics:** The compiler gains support for 128-bit atomic memory operations, relevant for lock-free data structures on 64-bit platforms.
- **AI Compute Extensions:** Initial support for ACE (AI Compute Extensions) has been added, targeting emerging accelerator and NPU workloads.
- **C++29 infrastructure:** Early parser and frontend groundwork for the C++29 standard is in place, continuing GCC's pattern of ahead-of-standard implementation.

## Why it matters
The GCC 17 freeze matters for Linux distribution maintainers and developers building performance-critical software, as the architecture-specific tuning improvements directly affect compiled output on current AMD Zen, Intel AVX10, and RISC-V hardware. For embedded and AI accelerator developers, the ACE support and SpacemiT X100 targeting expand GCC's reach into specialized compute domains where LLVM has been dominant.