Deploy Opencode Server to Serve Multiple Clients

Previously I mentioned that Opencode can actually be separated into Server and Client components. That means the Opencode Server🔗 can be deployed independently for other Opencode Clients to use. For example, a machine within a company can run local models or call APIs:

opencode serve --hostname 0.0.0.0 --port 4096 --mdns --mdns-domain opencode.local

# Ollama 與 opencode 有深度整合，設置模型很方便
# https://docs.ollama.com/integrations/opencode

Any client on the intranet can connect to http://opencode.local:4096 to operate, whether it’s a browser, TUI, or an app: https://opencode.ai/zht/download🔗

# In another terminal, attach the TUI to the running backend
opencode attach http://opencode.local:4096

Right now the server can host and serve its own models; per-client management isn’t available yet. There is an existing service called Zen🔗 you can check out, or LiteLLM🔗.