Issues

• Logits processors receive unverified draft tokens in input_ids when a draft_model (prompt lookup) is used

  abetlen/llama-cpp-python#2393 · Filter by author YaochengTong opened on Oct 28, 2026

• Concurrent Llama.embed() aborts the process (GGML_ASSERT), no exception raised

  abetlen/llama-cpp-python#2392 · Filter by author JiaY-77 opened on Oct 26, 2026

• Seven llama_sampler_init_* bindings admit keyword arguments that the ctypes function object silently drops

  abetlen/llama-cpp-python#2371 · Filter by author GreenFuze opened on Sep 22, 2026

• Long whitespace tokens detokenize to empty bytes, breaking tokenize/detokenize round-trip

  abetlen/llama-cpp-python#2362 · Filter by author gbjones opened on Sep 4, 2026

• [Bug] DFlash 2 speculative decoding crashes and fails in Python due to missing ctx_other and non-autoregressive graph execution

  abetlen/llama-cpp-python#2361 · Filter by author ilperev opened on Aug 20, 2026
  1

• uv add llama-cpp-python wheels fails for versions above 0.3.30

  abetlen/llama-cpp-python#2352 · Filter by author Karthik777 opened on Jul 31, 2026
  1

• Support sm100 and sm120 in CUDA pre‑built wheels

  abetlen/llama-cpp-python#2341 · Filter by author XuehaoSun opened on Jul 16, 2026

• Feature Request: SuperCompress prompt preprocessing for llama-cpp-python

  abetlen/llama-cpp-python#2326 · Filter by author arjunkshah opened on Jun 30, 2026

• Docs: consolidate build-from-source and GPU backend guide

  abetlen/llama-cpp-python#2314 · Filter by author cschanhniem opened on Jun 19, 2026

• Request for Gemma 3 4B & Gemma 4 Integration Support with llama-cpp-python

  abetlen/llama-cpp-python#2231 · Filter by author LeConsulat2 opened on May 25, 2026
  2

• Gemma 4 tool calls returned as raw native tokens in content instead of tool_calls

  abetlen/llama-cpp-python#2227 · Filter by author ThiruvarankanM opened on May 22, 2026

• Llava15ChatHandler - Cache Image Encoding

  abetlen/llama-cpp-python#2222 · Filter by author d-orm opened on May 17, 2026

• Asset seeder disabled 'resemble-perth' not found. watermarking will be unavailable.

  abetlen/llama-cpp-python#2219 · Filter by author andymoorey opened on May 16, 2026

• Llama.embed() calls LlamaBatch.add_sequence with old 3-arg signature; missing logits_array

  abetlen/llama-cpp-python#2211 · Filter by author emptyngton opened on May 9, 2026
  2

• Llama() silently accepts and discards embedding kwarg; .embed() then raises confusingly

  abetlen/llama-cpp-python#2210 · Filter by author emptyngton opened on May 9, 2026

• Can RotorQuant/TurboQuant and dflash support be added?

  abetlen/llama-cpp-python#2184 · Filter by author Sourajit1234 opened on Apr 16, 2026

• How to use "Gemma-4-E4B-it-heretic-GGUF" ?

  abetlen/llama-cpp-python#2175 · Filter by author TimmyHeart opened on Apr 6, 2026
  3

• Improve error messages

  abetlen/llama-cpp-python#2145 · Filter by author mausch opened on Mar 22, 2026

• feat: Update vendored llama.cpp to include ModernBERT support

  abetlen/llama-cpp-python#2144 · Filter by author clemlesne opened on Mar 16, 2026

• Multiple calls to create_chat_completion() fail with "llama_decode: failed to decode, ret = -1"

  abetlen/llama-cpp-python#2140 · Filter by author thisisayushg opened on Mar 14, 2026

• Improve installation DX: prebuilt wheels for 3.13/3.14/3.14t + declarative backend selection

  abetlen/llama-cpp-python#2136 · Filter by author clemlesne opened on Mar 7, 2026

• Pre-built CPU-only wheel for Windows (cp313) for version 0.3.16+ (Gemma 3 support)

  abetlen/llama-cpp-python#2130 · Filter by author mselcukozturk opened on Feb 24, 2026
  1

• Up to date llama.cpp wheel here (native libraries)

  abetlen/llama-cpp-python#2123 · Filter by author mdjou opened on Feb 13, 2026

• Pre-built manylinux wheel for llama_cpp_python (Linux, Python 3.10)

  abetlen/llama-cpp-python#2119 · Filter by author mrzeeshanahmed opened on Jan 30, 2026

• [Deployment Issue] Installation deadlock on Hugging Face Spaces (CPU): Wheels fail (musl/glibc mismatch) & Source builds timeout

  abetlen/llama-cpp-python#2118 · Filter by author Po-Hsuan-Huang opened on Jan 26, 2026