Problemsby GlyphTrouble with llama.cpp misc
22 posts from 22 people18d activeSteady, +1 this week
Posts per day
The posts behind it
Misc. bug: allOf with non-object types generates an empty object grammar
...ends with `_build_object_rule()`, so anything else drops through to an empty object Workaround for now is flattening the allOf before it reaches llama-server, E.G into `{"type":"integer","minimum":0,"maximum":100}` Possibly related - #15931 (nested object allOf, fixed) and...
Misc. bug: Code incomplete looks too similar to Copy code
### Name and Version version: 0.8.0-dev (build 12014, commit aa588a346) built with Clang 20.1.8 for Windows x86_64 ### Operating systems Windows ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description & steps to reproduce If...
Misc. bug: router memory measurement leaves out the draft model
### Bug With `--models-memory-margin` (added in #21231), the router measures each model's memory (`min` and `max` in `GET /models`) in a short-lived child process. That child loads only the main model (`common_measure_memory` → `common_fit_measure` in `common/fit.cpp`), so the...
Misc. bug: Speculative Decode Performance Drop
### Name and Version version: 0.7.0-dev (build 11870, commit 74bca3950) built with GNU 13.3.0 for Linux aarch64 ### Operating systems Linux ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description & steps to reproduce Tested...
Misc. bug: Images don't get re-tagged properly
### Name and Version Potentially every version starting from 0.6 I guess ### Operating systems _No response_ ### Which llama.cpp modules do you know to be affected? Other (Please specify in the next section) ### Command line ### Problem description & steps to reproduce Currently...
Misc. bug: llama_get_logits_ith() reorders outputs in place, so concurrent reads race (recurrent/hybrid models)
### Name and Version v0.7.0 (620d222), Metal build. The code is the same on master as of 2280f0fd1. ### Operating systems Mac ### Which llama.cpp modules do you know to be affected? libllama (core library) ### Problem description & steps to reproduce `llama_get_logits_ith()`...
Misc. bug: MUL_MAT: wrong results, exceeds tolerance; gfx1200 HIP/ROCm; f16 + deterministic.
...across runs with different random inputs. So it's a systematic precision problem, not a flaky race. Not claimed: No root cause. Repro reports 0/1 tests passed. Build Command: cmake -S . -B build-hip -DGGML_HIP=ON -DGPU_TARGETS=gfx1200 \ -DLLAMA_BUILD_TESTS=ON...
Misc. bug: docs mention --spec-draft-conf-min, but that flag doesn't exist
### Name and Version version: 0.7.0-dev (build 11868, commit 78a8ae14b) built with Clang 20.1.8 for Windows x86_64 ### Operating systems Windows ### Which llama.cpp modules do you know to be affected? Documentation/Github ### Command line ### Problem description & steps to...
Misc. bug: Sarvam 30B MoE converter labels byte-fallback tokens <0x00>–<0xFF> NORMAL instead of BYTE
### Name and Version version: 0.7.0-dev (build 11731, commit 42b70032b) built with Clang 20.1.8 for Windows x86_64 ### Operating systems Windows ### Which llama.cpp modules do you know to be affected? Python/Bash scripts ### Command line ### Problem description & steps to...
Misc. bug: Some browsers have web search tools, other don't have them.
### Name and Version version: 0.5.0-dev (build 11345, commit a868c3e3c) built with MSVC 19.51.36248.0 for Windows AMD64 ### Operating systems Windows ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description & steps to...
Misc. bug: Smart App Control blocks DLLs in official Windows Vulkan bundles (b11429 / b11535)
### Name and Version llama-server from the official Windows x64 Vulkan bundles: - llama-b11429-bin-win-vulkan-x64.zip - llama-b11535-bin-win-vulkan-x64.zip The older b9209 Vulkan runtime worked in tests on the same PC. ### Operating systems Windows ### Which llama.cpp modules do...
Misc. bug: Vulkan: FLASH_ATTN_EXT wrong results on MoltenVK 1.4.2 + AMD RDNA1 when head size is not a multiple of subgroup size (Apple+AMD shuffle disable from #15846 still required)
...currently force-disables `subgroup_shuffle` on Apple + AMD: I tested whether this workaround is still necessary on MoltenVK 1.4.2 by commenting out **only** the `subgroup_shuffle = false` line (the `subgroup_arithmetic` disable was left in place to keep the experiment isolated),...
Misc. bug: PLaMo3: tokenization results differ from the reference
### Name and Version version: 0.5.0-dev (build 11416, commit 2ed93db47) built with Clang 22.1.8 for Linux x86_64 ### Operating systems Linux ### Which llama.cpp modules do you know to be affected? Other (Please specify in the next section) ### Command line ### Problem...
Misc. bug: CPU flash attention produces NaN when the first key has an -inf score
### Name and Version commit: 50569eb87df530daff11afda229ceb9ab8e6cae8 ggml version: 0.26.0 Built from source on MacOS Arm64 clang ### Operating systems Mac, Linux ### Which llama.cpp modules do you know to be affected? libllama (core library) ### Command line ### Problem...
Misc. bug: Homebrew's llama.cpp 0.5.0 has no web UI
### Name and Version Homebrew llama.cpp 0.5.0 ### Operating systems Mac ### Which llama.cpp modules do you know to be affected? llama-server ### Problem description & steps to reproduce The Homebrew build of llama.cpp `0.5.0` ships without the web UI. `llama serve` answers `/`...
Misc. bug: 76a5bc86d1bdfae96feccdc7a41fea535e792e6e breaks symlinked cache dirs for RPC
...to a symlink to a directory and try to start rpc server. It will fail with a message unable to create .cache/llama.cpp/rpc ### First Bad Commit 76a5bc86d1bdfae96feccdc7a41fea535e792e6e ### Relevant log output Logs
Misc. bug: Ling 3.0 parser ignores response_format / json_schema
### Name and Version version: 0.5.0-dev (build 11223, commit 4da633776) built with GNU 14.2.0 for Linux x86_64 docker image ghcr.io/ggml-org/llama.cpp:full-cuda-b11223, CUDA. The same code is in master at 31385c9ce. ### Operating systems _No response_ ### Which llama.cpp modules...
Misc. bug: Releases page appears newly busted
### Name and Version Github Releases page ### Operating systems _No response_ ### Which llama.cpp modules do you know to be affected? _No response_ ### Command line ### Problem description & steps to reproduce Releases since...
Misc. bug: server: (router) --api-key is not forwarded to child instances while --api-key-file is
### Name and Version version: 0.4.0-dev (build 158, commit 311d4211b) ### Operating systems Mac ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description & steps to reproduce In router mode, --api-key and --api-key-file are...
Misc. bug: Severe drop in prompt processing speed after b10780 on Vulkan, RDNA3
### Name and Version ❯ llama-server --version version: 0.3.0-dev (build 10781, commit unknown) built with GNU 16.2.1 for Linux x86_64 ### Operating systems Linux ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description & steps...
Misc. bug: Race condition in router scheduler during concurrent cold-start with --models-max 1
### Name and Version > llama-server --version version: 0.4.0-dev (build 10872, commit b31b71f3a) built with GNU 15.2.0 for Linux x86_64 ### Operating systems Linux ### Which llama.cpp modules do you know to be affected? llama-server ### Command line ### Problem description &...
Misc. bug: Llama.cpp localhost shows the ui even when llama.cpp isnt running.
### Name and Version Latest version ### Operating systems Windows ### Which llama.cpp modules do you know to be affected? Other (Please specify in the next section) ### Command line ### Problem description & steps to reproduce Just run llama.cpp enough and for some reason it'll...
Companies and products named
About this problem
Evidence
22 posts from 22 people in 1 place, about 2 a week over 63 days. Mostly on ggml-org/llama.cpp. Tools named alongside: GitHub, NVIDIA, Docker, Hugging Face.
Frustration Frustration 0 of 3· Seen on GitHub issues
Summary, model-written
Llama.cpp users experience unexpected runtime failures, router mode API key issues, and server bugs across Windows and Linux platforms.
How it was grouped
Posts about llama.cpp with no known issue, sharing the word "misc". First post Sep 5, 2026, latest Nov 6, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
Other llama.cpp problems
Running llama.cpp locally hits memory and speed limits
45 posts from 41 people on GitHub issues, Reddit.
Experiments on llama.cpp are hard to reproduce and track
2 posts from 2 people on GitHub issues. Unverified.
llama.cpp users keep filing bugs
Top post: "llama-ui : unable to open reasoning level selection menu on desktop", one of 13 posts from 13 people on GitHub issues, Reddit.
llama.cpp upgrades break working setups
6 posts from 6 people on GitHub issues.
Trouble with llama.cpp compile
Top post: "Compile bug: rocm-6.3.4, no matching function for call to group_mask_warp_inplace", one of 5 posts from 5 people on GitHub issues.
Fine-tuning llama.cpp is costly and unpredictable
4 posts from 4 people on GitHub issues, Hacker News.
llama.cpp runs out of context on real work
4 posts from 4 people on GitHub issues.
llama.cpp lock-in makes switching costly
3 posts from 3 people on GitHub issues.
llama.cpp runs out of memory
3 posts from 3 people on GitHub issues.
Trouble with llama.cpp flash
Top post: "Feature Request: Support GLM5.3 (flash)", one of 2 posts from 2 people on GitHub issues, Reddit. Unverified.
llama.cpp fails to return valid structured output
2 posts from 2 people on GitHub issues. Unverified.
Trouble with llama.cpp builds
Top post: "Feature Request: update ROCm release builds 10.0.0 -> 10.1.0", one of 2 posts from 2 people on GitHub issues. Unverified.
History