Misc. bug:

[jack-liew]

Name and Version

llama.cpp b9701

When I run this command:
llama-server --models-dir "D:\models\ggufs\Models" --host 127.0.0.1 --port 10001...

The console prompts:
srv llama_server: router server is listening on http://127.0.0.1:10001.

However, no model list JSON is returned when accessing http://127.0.0.1:10001/v1/models.

In the console output of llama-server, I can see:
srv proxy_reques: proxying request to model Jackrong/Qwopus3.5-9B-Coder-MTP-GGUF:Q6_K on port 64239.

and I can get the JSON of all models by accessing http://127.0.0.1:64239/v1/models. However, this port 64239 can not fixed, and it only returns the JSON for a single model. What I need is to access the fixed port 10001/v1/models to get the list of all models.

Strangely, the version of llama.cpp released the day before yesterday could still return the JSON of all models by accessing http://127.0.0.1:64239/v1/models.

Operating systems

Windows

Which llama.cpp modules do you know to be affected?

llama-server

Command line

llama-server --models-dir "D:\models\ggufs\Models" --host 127.0.0.1 --port 10001

Problem description & steps to reproduce

run : llama-server --models-dir "D:\models\ggufs\Models" --host 127.0.0.1 --port 10001

accessing : http://127.0.0.1:10001/v1/models

Thanks for great job!