AI Briefing
KO

llama.cpp Updates Vulkan Backend and Adds New Model Support

·2026.06.24 18:37

Key point

llama.cpp has rolled out an update including Vulkan backend performance improvements and support for new models such as Granite and LFM2.5.

Details

The latest llama.cpp update introduces the following key changes.

New Model Support

  • Support for granite-speech-4.1-2b-plus
  • Support for LFM2.5-ColBERT-350M and LFM2.5-Embedding-350M

Vulkan Backend Improvements

  • Added CONV_3D support and GET_ROWS_BACK functionality
  • Added backend test support for various operations including SQR, SQRT, SIN, COS, CLAMP, LEAKY_RELU, NORM
  • Applied Bias before Softmax to prevent overflow
  • Vulkan-related optimizations and bug fixes

Other Notes

  • Includes UI logo replacement, navigation cleanup, and mobile UI/UX improvements.
llama.cppvulkaninference-enginegranite-speechlfm2.5
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditNew Feature Update for ik_llama.cppHuggingFace BlogLFM2.5 Inference Accelerated 3.2x with DSparkRedditInference Engine Support Status for Ling-3.0-flashGeekNewsUnderdog Releases 'Husky', a Model-Specific Inference Engine Up to 4.5x Faster than MLXTLDR AILLM Can Control Host Machine by Exploiting Inference Engine VulnerabilitiesHacker NewsLLMs Can Exploit Inference Engine Vulnerabilities to Control Host Machines