Name and Version
./llama-server --version
version: 9501 (65ef50a0a)
built with GNU 15.2.0 for Linux x86_64
Note: while I was writing this issue, it seems the version 9501 was pulled back from releases
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
No response
Command line
Problem description & steps to reproduce
Symptom
Connecting to the web ui via the browser (http://localhost:10000/) shows:
{"error":{"message":"File Not Found","type":"not_found_error","code":404}}podman run -d \
--pull=newer \
--name gemma4_12B \
--device /dev/dri/ \
--network pasta:--map-guest-addr,10.0.0.1 -p 10000:10000 \
-v /home/<redacted>/.cache/llama.cpp/:/models \
-v /home/<redacted>/.cache/llama.cpp:/root/.cache/llama.cpp \
-v /home/<redacted>/.cache/huggingface/hub/:/root/.cache/huggingface/hub/ \
ghcr.io/ggml-org/llama.cpp:server-vulkan \
-hf unsloth/gemma-4-12b-it-GGUF:Q6_K \
--port 10000 \
--host 0.0.0.0 \
--ctx-size 32768 \
--batch-size 512 \
--temp 1.0 \
--reasoning on
the model does not matter
Workaround
Download the web UI assets once and mount them:
mkdir -p ~/.cache/llama.cpp/ui/dist
for f in index.html bundle.js bundle.css loading.html; do
curl -sfL -o ~/.cache/llama.cpp/ui/dist/$f \
"https://huggingface.co/buckets/ggml-org/llama-ui/resolve/latest/$f?download=true"
doneThen add to the podman/docker run command:
-v ~/.cache/llama.cpp/ui/dist:/ui:ro --path /ui
What I run to fix the issue (I have a RX 9070XT so YMMV)
podman run -d \
--pull=newer \
--name gemma4_12B \
--device /dev/dri/ \
--network pasta:--map-guest-addr,10.0.0.1 -p 10000:10000 \
-v /home/<redacted>/.cache/llama.cpp/:/models \
-v /home/<redacted>/.cache/llama.cpp:/root/.cache/llama.cpp \
-v /home/<redacted>/.cache/huggingface/hub/:/root/.cache/huggingface/hub/ \
-v /home/<redacted>/.cache/llama.cpp/ui/dist:/ui:ro ghcr.io/ggml-org/llama.cpp:server-vulkan \
-hf unsloth/gemma-4-12b-it-GGUF:Q6_K \
--port 10000 \
--host 0.0.0.0 \
--ctx-size 32768 \
--batch-size 512 \
--temp 1.0 \
--reasoning on \
--path /uiFirst Bad Commit
Probably related to #22937
Relevant log output
Name and Version
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
No response
Command line
Problem description & steps to reproduce
Symptom
Connecting to the web ui via the browser (http://localhost:10000/) shows:
{"error":{"message":"File Not Found","type":"not_found_error","code":404}}podman run -d \ --pull=newer \ --name gemma4_12B \ --device /dev/dri/ \ --network pasta:--map-guest-addr,10.0.0.1 -p 10000:10000 \ -v /home/<redacted>/.cache/llama.cpp/:/models \ -v /home/<redacted>/.cache/llama.cpp:/root/.cache/llama.cpp \ -v /home/<redacted>/.cache/huggingface/hub/:/root/.cache/huggingface/hub/ \ ghcr.io/ggml-org/llama.cpp:server-vulkan \ -hf unsloth/gemma-4-12b-it-GGUF:Q6_K \ --port 10000 \ --host 0.0.0.0 \ --ctx-size 32768 \ --batch-size 512 \ --temp 1.0 \ --reasoning onWorkaround
Download the web UI assets once and mount them:
Then add to the podman/docker run command:
What I run to fix the issue (I have a RX 9070XT so YMMV)
podman run -d \ --pull=newer \ --name gemma4_12B \ --device /dev/dri/ \ --network pasta:--map-guest-addr,10.0.0.1 -p 10000:10000 \ -v /home/<redacted>/.cache/llama.cpp/:/models \ -v /home/<redacted>/.cache/llama.cpp:/root/.cache/llama.cpp \ -v /home/<redacted>/.cache/huggingface/hub/:/root/.cache/huggingface/hub/ \ -v /home/<redacted>/.cache/llama.cpp/ui/dist:/ui:ro ghcr.io/ggml-org/llama.cpp:server-vulkan \ -hf unsloth/gemma-4-12b-it-GGUF:Q6_K \ --port 10000 \ --host 0.0.0.0 \ --ctx-size 32768 \ --batch-size 512 \ --temp 1.0 \ --reasoning on \ --path /uiFirst Bad Commit
Probably related to #22937
Relevant log output