feat(version): add server version endpoint and client version display

Author: ilopezluna

.github/workflows/release.yml

@@ -137,6 +137,7 @@ jobs:
           platforms: linux/amd64, linux/arm64
           build-args: |
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -152,6 +153,7 @@ jobs:
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
             "LLAMA_SERVER_VARIANT=cuda"
             "BASE_IMAGE=nvidia/cuda:12.9.0-runtime-ubuntu24.04"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -170,6 +172,7 @@ jobs:
             "VLLM_VERSION=${{ inputs.vllmVersion }}"
             "VLLM_CUDA_VERSION=cu130"
             "VLLM_PYTHON_TAG=cp38-abi3"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -186,6 +189,7 @@ jobs:
             "LLAMA_SERVER_VARIANT=cuda"
             "BASE_IMAGE=nvidia/cuda:12.9.0-runtime-ubuntu24.04"
             "SGLANG_VERSION=${{ inputs.sglangVersion }}"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -199,6 +203,7 @@ jobs:
           platforms: linux/amd64, linux/arm64
           build-args: |
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -214,6 +219,7 @@ jobs:
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
             "LLAMA_SERVER_VARIANT=rocm"
             "BASE_IMAGE=rocm/dev-ubuntu-22.04"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -230,6 +236,7 @@ jobs:
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
             "LLAMA_SERVER_VARIANT=musa"
             "BASE_IMAGE=mthreads/musa:rc4.3.0-runtime-ubuntu22.04-amd64"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max
@@ -246,6 +253,7 @@ jobs:
             "LLAMA_SERVER_VERSION=${{ inputs.llamaServerVersion }}"
             "LLAMA_SERVER_VARIANT=cann"
             "BASE_IMAGE=ascendai/cann:8.2.rc2-910b-ubuntu22.04-py3.11"
+            "VERSION=${{ inputs.releaseTag }}"
           push: true
           sbom: true
           provenance: mode=max

Dockerfile

@@ -9,8 +9,12 @@ ARG LLAMA_BINARY_PATH=/com.docker.llama-server.native.linux.${LLAMA_SERVER_VARIA
 # use 22.04 for gpu variants to match ROCm/CUDA base images
 ARG BASE_IMAGE=ubuntu:26.04
 
+ARG VERSION=dev
+
 FROM docker.io/library/golang:${GO_VERSION}-bookworm AS builder
 
+ARG VERSION
+
 # Install git for go mod download if needed
 RUN apt-get update && apt-get install -y --no-install-recommends git && rm -rf /var/lib/apt/lists/*
 
@@ -30,13 +34,14 @@ COPY --link . .
 # Build the Go binary (static build)
 RUN --mount=type=cache,target=/go/pkg/mod \
     --mount=type=cache,target=/root/.cache/go-build \
-    CGO_ENABLED=1 GOOS=linux go build -ldflags="-s -w" -o model-runner .
+    CGO_ENABLED=1 GOOS=linux go build -ldflags="-s -w -X main.Version=${VERSION}" -o model-runner .
 
 # Build the Go binary for SGLang (without vLLM)
 FROM builder AS builder-sglang
+ARG VERSION
 RUN --mount=type=cache,target=/go/pkg/mod \
     --mount=type=cache,target=/root/.cache/go-build \
-    CGO_ENABLED=1 GOOS=linux go build -tags=novllm -ldflags="-s -w" -o model-runner .
+    CGO_ENABLED=1 GOOS=linux go build -tags=novllm -ldflags="-s -w -X main.Version=${VERSION}" -o model-runner .
 
 # --- Get llama.cpp binary ---
 FROM docker/docker-model-backend-llamacpp:${LLAMA_SERVER_VERSION}-${LLAMA_SERVER_VARIANT} AS llama-server

Makefile

@@ -32,7 +32,7 @@ BUILD_DMR ?= 1
 
 # Build the Go application
 build:
-	CGO_ENABLED=1 go build -ldflags="-s -w" -o $(APP_NAME) .
+	CGO_ENABLED=1 go build -ldflags="-s -w -X main.Version=$(shell git describe --tags --always --dirty --match 'v*')" -o $(APP_NAME) .
 
 # Run the application locally
 run: build

cmd/cli/commands/version.go

@@ -1,6 +1,8 @@
 package commands
 
 import (
+	"runtime"
+
 	"github.com/docker/model-runner/cmd/cli/commands/completion"
 	"github.com/docker/model-runner/cmd/cli/desktop"
 	"github.com/spf13/cobra"
@@ -11,8 +13,24 @@ func newVersionCmd() *cobra.Command {
 		Use:   "version",
 		Short: "Show the Docker Model Runner version",
 		Run: func(cmd *cobra.Command, args []string) {
-			cmd.Printf("Docker Model Runner version %s\n", desktop.Version)
-			cmd.Printf("Docker Engine Kind: %s\n", modelRunner.EngineKind())
+			cmd.Println("Client:")
+			cmd.Printf(" Version:    %s\n", desktop.Version)
+			cmd.Printf(" OS/Arch:    %s/%s\n", runtime.GOOS, runtime.GOARCH)
+
+			cmd.Println()
+			cmd.Println("Server:")
+			if desktopClient == nil {
+				cmd.Println(" Version:    (not reachable)")
+				cmd.Println(" Engine:     (not reachable)")
+				return
+			}
+			sv, err := desktopClient.ServerVersion()
+			if err != nil {
+				cmd.Println(" Version:    (not reachable)")
+			} else {
+				cmd.Printf(" Version:    %s\n", sv.Version)
+			}
+			cmd.Printf(" Engine:     %s\n", modelRunner.EngineKind())
 		},
 		ValidArgsFunction: completion.NoComplete,
 	}

cmd/cli/desktop/desktop.go

@@ -694,6 +694,34 @@ func (c *Client) Remove(modelArgs []string, force bool) (string, error) {
 	return modelRemoved, nil
 }
 
+type ServerVersionResponse struct {
+	Version string `json:"version"`
+}
+
+func (c *Client) ServerVersion() (ServerVersionResponse, error) {
+	resp, err := c.doRequest(http.MethodGet, "/version", nil)
+	if err != nil {
+		return ServerVersionResponse{}, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return ServerVersionResponse{}, fmt.Errorf("failed to get server version: %s", resp.Status)
+	}
+
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return ServerVersionResponse{}, fmt.Errorf("failed to read response body: %w", err)
+	}
+
+	var version ServerVersionResponse
+	if err := json.Unmarshal(body, &version); err != nil {
+		return ServerVersionResponse{}, fmt.Errorf("failed to unmarshal response body: %w", err)
+	}
+
+	return version, nil
+}
+
 // BackendStatus to be imported from docker/model-runner when https://github.com/docker/model-runner/pull/42 is merged.
 type BackendStatus struct {
 	BackendName string               `json:"backend_name"`

main.go

@@ -3,6 +3,7 @@ package main
 import (
 	"context"
 	"crypto/tls"
+	"encoding/json"
 	"net"
 	"net/http"
 	"os"
@@ -264,6 +265,12 @@ func main() {
 	anthropicHandler := anthropic.NewHandler(log, schedulerHTTP, nil, modelManager)
 	router.Handle(anthropic.APIPrefix+"/", anthropicHandler)
 
+	// Register /version endpoint
+	router.HandleFunc("/version", func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "application/json")
+		_ = json.NewEncoder(w).Encode(map[string]string{"version": Version})
+	})
+
 	// Register root handler LAST - it will only catch exact "/" requests that don't match other patterns
 	router.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
 		// Only respond to exact root path

version.go

@@ -0,0 +1,3 @@
+package main
+
+var Version = "dev"