Search results

Bug: qwen 3.5 full prompt bug

Status: Open.

#1319 In ikawrakow/ik_llama.cpp;

· gopinath87607 opened on Feb 25, 2026

Feature Request: LFM2-24B-A2B and LFM2-8B-A1B(lfm2moe architecture) support

enhancementNew feature or requestNew feature or request

Status: Open.

#1317 In ikawrakow/ik_llama.cpp;

· hardWorker254 opened on Feb 24, 2026

Bug: llama-quantize failing on qwen3next low BPW quant types.

Status: Open.

#1316 In ikawrakow/ik_llama.cpp;

· ubergarm opened on Feb 24, 2026

Bug: Crash after generation with layer split

Status: Open.

#1312 In ikawrakow/ik_llama.cpp;

· Geechan opened on Feb 24, 2026

Feature Request: Pre-calculate required VRAM for everything instead of just model weight

enhancementNew feature or requestNew feature or request

Status: Open.

#1302 In ikawrakow/ik_llama.cpp;

· chulucninh09 opened on Feb 22, 2026

Bug: Kimi K2.5 sometimes generates garbage output at long context

Status: Open.

#1298 In ikawrakow/ik_llama.cpp;

· Lissanro opened on Feb 21, 2026

Bug: coredump at llama_grammar_accept_token()

Status: Open.

#1297 In ikawrakow/ik_llama.cpp;

· alex1284B opened on Feb 21, 2026

Bug: Continue dev returns blank replies from openAI endpoint.

Status: Open.

#1273 In ikawrakow/ik_llama.cpp;

· Ph0rk0z opened on Feb 16, 2026

Garbage output when using --cpu-moe with Step-3.5-Flash or Minimax-M2.1

Status: Open.

#1256 In ikawrakow/ik_llama.cpp;

· buckleup1987 opened on Feb 8, 2026

Bug: gptoss 120b crashed

Status: Open.

#1234 In ikawrakow/ik_llama.cpp;

· calvin2021y opened on Feb 5, 2026

Issues with string ban

Status: Open.

#1233 In ikawrakow/ik_llama.cpp;

· SneedwareInc opened on Feb 5, 2026

Bug: (Probably OOM) Crash after several identical queries to gpt-oss-20b on dual RTX 3090 with --sm graph

Status: Open.

#1232 In ikawrakow/ik_llama.cpp;

· olobarius opened on Feb 4, 2026