b10215

ggml-org/llama.cppb10215Jul 31, 2026by github-actions[bot]same day as b10214

What changed since b10214

low risk

b10215 adds a Vulkan driver version check for Windows Intel GPUs to prevent crashes, replacing the previous Intel crash guard.

New features

  • Vulkan backend now checks the Windows Intel GPU driver version and requires 32.0.101.8860 or newer for the affected path.

Improvements

  • Driver version detection was refactored to compare numeric driverVersion values instead of parsing strings.
  • The check permits GPUs beyond Xe2+ rather than restricting to a single generation.
  • The blanket Intel crash guard was removed now that the underlying driver bug is fixed in newer drivers.

Bug fixes

  • Resolves crashes observed on Windows Intel GPUs running Vulkan with drivers older than 32.0.101.8860.

Upgrade notes

No config or API changes are required. Windows Intel GPU users on Vulkan should update their graphics driver to 32.0.101.8860 or newer, since older drivers are now gated by the version check. Other backends and platforms are unaffected.

Analyzed Nov 1, 2026

Upgrading?

Release notes summary

Fixes a crash on Windows Intel GPUs by introducing a driver version check to ensure compatibility with driver version 32.0.101.8860 and above.

Key Highlights

  • Windows Intel GPU crash mitigation
  • Vulkan driver version check implementation
  • Refactored driver version detection logic

Release Notes (excerpt)

<details open> vulkan: Introduce driver version check for Windows Intel GPU to mitigate crashing (#25192) * Removed crash guard for Intel Crash fixed from driver 32.0.101.8860 * Added driver version check for windows * Change to convert from driverVersion rather than string * No need to use signed * Refactor * allow GPU other than Xe2+ * adjusted function body position </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories