Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

`docker model` CLI detects "Docker Engine" on Docker Desktop for Windows and tries to start standalone runner (port 12434 conflict)

Open
  • Is something wrong?
#1,077 1 comment 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
45/100
Issue type
Bug
Clarity
Mostly clear
Activity status
Active
Tech stack
docker, go
Domain
cli

Research direction

Start with the bundled plugin at C:\Program Files\Docker\cli-plugins\docker-model.exe and reproduce the issue with docker model status against the desktop-linux context. Check how the plugin distinguishes Docker Desktop's runner from a standalone runner and selects the runner image, using the working http://localhost:12434/engines/v1/models endpoint as a reference. Done means the model commands use the available Desktop runner without the port conflict or an inappropriate CUDA image.

Written by the indexing model from the issue text.

Description

Environment

  • Windows, Docker Desktop 4.93.0 (240920), context desktop-linux
  • docker model version: Client v1.2.6 / Server v1.2.8, Engine: Docker Engine
  • Plugin path (from docker info): C:\Program Files\Docker\cli-plugins\docker-model.exe
  • CPU: Intel (no NVIDIA GPU)

Problem
Every docker model command (status, ls, list, run) pulls docker/model-runner:latest-cuda and tries to start a standalone docker-model-runner container, which fails because Docker Desktop's Model Runner already listens on 127.0.0.1:12434 (owned by com.docker.backend.exe):

unable to initialize standalone model runner: unable to initialize standalone model runner container: failed to start container docker-model-runner: Error response from daemon: ports are not available: exposing port TCP 127.0.0.1:12434 -> 127.0.0.1:0: listen tcp4 127.0.0.1:12434: bind: Only one usage of each socket address (protocol/network address/port) is normally permitted.

Evidence the Desktop runner works
curl http://localhost:12434/engines/v1/models returns {"object":"list","data":[]}.

Already tried

  • Machine reboot
  • docker desktop disable model-runner + docker desktop enable model-runner --tcp=12434
  • With the Desktop runner disabled, the standalone runner fails binding 172.17.0.1:12434 ("can't bind on the specified endpoint")
  • No MODEL_RUNNER_HOST / DOCKER_HOST env vars set

Notable

  1. CLI reports Engine: Docker Engine while docker version reports Server: Docker Desktop 4.93.0 (240920).
  2. CLI pulls the latest-cuda image on a machine without an NVIDIA GPU.
  3. Client/server version mismatch (1.2.6 vs 1.2.8).

Possibly related: #377 (closed). The fix there was removing a stale plugin from ~/.docker/cli-plugins, but here the resolved plugin is the one bundled with Docker Desktop.

Dominant language
Go
Stars
656
Forks
156
Avg merge
3d 10h
Merged PRs (30d)
8

Getting set up

  • Ships a Dockerfile or Docker Compose file
  • No pull request template
  • No contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from docker/model-runner

All issues in docker/model-runner

Similar issues

More Go issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.