Discussions

• 99 You must be logged in to vote

  🙌

  guide : using llama-ui — the new WebUI of llama.cpp

  ggerganov started Nov 2, 2025 in Show and tell

  

  

  

  

  253
• 1 You must be logged in to vote

  🙏

  Struggling to get Vulkan to work for 285hx iGPU

  StudioNirin asked Jul 13, 2026 in Q&A · Unanswered

  

  

  0
• 2 You must be logged in to vote

  🙏

  Release Cycle of llama.cpp

  yavornenov-ost asked Sep 19, 2025 in Q&A · Answered

  

  

  

  

  41
• 2 You must be logged in to vote

  🙏

  Current status of Intel Arc GPUs for llama.cpp

  ky438 asked Mar 25, 2025 in Q&A · Unanswered

  

  

  

  

  25
• 6 You must be logged in to vote

  💬

  Mmap faster than direct I/O for MoE models

  qdacsvx started Jan 11, 2026 in General

  

  

  

  

  3
• 2 You must be logged in to vote

  🙌

  Gemma4 26B on Radeon 780M iGPU via llama.cpp Vulkan — ~23–25 tok/s

  l1v1ngth3dr34m started Jun 6, 2026 in Show and tell

  

  

  1
• 1 You must be logged in to vote

  🙌

  Per-expert imatrix stats reveal domain-dependent MoE routing — used them to prune 25% of OLMoE's experts (tool + measurements)

  lucafulgenzi started Jul 13, 2026 in Show and tell

  

  

  0
• 2 You must be logged in to vote

  🙏

  What happened to the detailed starting/loading output?

  segmond asked May 30, 2026 in Q&A · Unanswered

  

  

  

  

  8
• 1 You must be logged in to vote

  💡

  Feature Request: OSCAR/Q2 support

  giveen started Jul 13, 2026 in Ideas · Closed

  

  

  1
• 4 You must be logged in to vote

  🙏

  How to use --tools all

  alan-l asked Apr 19, 2026 in Q&A · Unanswered

  

  

  

  

  17
• 2 You must be logged in to vote

  🙏

  Now the cli need access to ports, WTF?

  Mainframework asked Jul 9, 2026 in Q&A · Closed · Unanswered

  

  

  2
• 7 You must be logged in to vote

  💬

  Insufficient documentation for -ot and --override-tensor flag

  Juahyori started Apr 28, 2025 in General

  

  

  

  

  10
• 4 You must be logged in to vote

  🙌

  E8 lattice KV cache: 2-bit that preserves retrieval (30/30 vs TurboQuant 0/30)

  jagmarques started Jul 6, 2026 in Show and tell

  

  

  2
• 1 You must be logged in to vote

  💬

  SYCL Backend Removal of MKL Dependency

  MordragT started Jul 11, 2026 in General

  

  

  0
• 1 You must be logged in to vote

  🙌

  Metal Q8_0 KV quantization performance: GPT 5.6 Sol seems to have implemented some massive improvements

  tarruda started Jul 11, 2026 in Show and tell

  

  

  2
• 1 You must be logged in to vote

  💡

  WebUI Code Editor / Canvas / Artifacts

  Danmoreng started Apr 10, 2026 in Ideas

  

  

  

  

  3
• 43 You must be logged in to vote

  📣

  Add GPU support to ggml

  enhancement New feature or request help wanted Needs help from the community hardware Hardware related research 🔬

  ggerganov announced Apr 12, 2023 in Announcements

  

  

  

  

  210
• 1 You must be logged in to vote

  🙏

  Efficacious Benchmark - llama.cpp vs vLLM

  d-shehu asked Jul 9, 2026 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  💬

  Added Directx backend to a separate copy. Slow but works. under testing.

  Maxritz started Jul 10, 2026 in General

  

  

  0
• 2 You must be logged in to vote

  💬

  Pascal support

  shockme started Jul 10, 2026 in General

  

  

  0
• 200 You must be logged in to vote

  📣

  Inference at the edge

  ggerganov announced Mar 16, 2023 in Announcements

  

  

  

  

  65
• 5 You must be logged in to vote

  🙌

  Performance of llama.cpp on Intel GPU with SYCL backend

  performance Speed related topics

  arthw started May 19, 2026 in Show and tell

  

  

  

  

  70
• 5 You must be logged in to vote

  💬

  [Windows 11] AMD Adrenalin driver (32.0.31007.5012) causes R9700 VRAM to offload to system RAM on idle - crash with 32GB RAM

  realsheox-max started May 20, 2026 in General

  

  

  

  

  2
• 41 You must be logged in to vote

  💬

  Performance of llama.cpp with Vulkan

  performance Speed related topics need feedback Testing and feedback with results are needed

  netrunnereve started Dec 18, 2024 in General

  

  

  

  

  750
• 2 You must be logged in to vote

  💡

  server UI: --tools Agent from Markdown and security aspects

  farhi started Jun 20, 2026 in Ideas

  

  

  

  

  2