[Unit]
Description=A Llama CPP Server Running a Reasoning Model

[Container]
# Shared AI internal pod without internet access
Pod=ai-internal.pod

# Image is built locally via podman build
# latest-mtp is for mtp testing
# latest is main branch
Image=localhost/llama-cpp-vulkan:latest

# Downloaded models volume
Volume=/home/ai/models/text:/models:z

# GPU Device
AddDevice=/dev/kfd
AddDevice=/dev/dri

# Server command
Exec=--port 8000 \
    -c 262144 \
    -n 32768 \
    --temp 0.6 \
    --top-p 0.95 \
    --top-k 20 \
    --min-p 0.0 \
    --presence-penalty 0.0 \
    --repeat-penalty 1.0 \
    --reasoning-budget 4096 \
    --perf \
    --n-gpu-layers all \
    -fa on \
    --spec-type draft-mtp --spec-draft-n-max 2 \
    --jinja \
    --chat-template-kwargs '{"preserve_thinking": true}' \
    --chat-template-file /models/qwen3.6-35b-a3b-mtp/chat_template.jinja \
    -m /models/qwen3.6-35b-a3b-mtp/Qwen3.6-35B-A3B-UD-Q5_K_M.gguf \
    --mmproj /models/qwen3.6-35b-a3b-mtp/mmproj-F16.gguf \
    --alias turbo

# Health Check
HealthCmd=CMD-SHELL curl --fail http://127.0.0.1:8000/health || exit 1
HealthInterval=10s
HealthRetries=3
HealthStartPeriod=10s
HealthTimeout=30s
HealthOnFailure=kill

[Service]
Restart=always
# Extend Timeout to allow time to pull the image
TimeoutStartSec=900

[Install]
# Start by default on boot
WantedBy=multi-user.target default.target