opencode zen is a curated, “ready-to-use” gateway of tested models optimized for coding agents. It integrates as an optional provider within opencode, offering pay-as-you-go access to top models (e.g. GPT-5, Claude, Grok) with no lock-in.
I've been an opencode user for a little while now and so far have really enjoyed the zen experience over the last week. I haven't been diligent enough about subscribing to all the model providers out there to figure out which one I should be using for a given task, and zen has given me a single-entry-point solution to enable that sort of experimentation. I also appreciate that the models that are available in this list are coming recommended for coding with bespoke handling to squeeze the best juice out of each one.
The models available today include those listed below, plus a few stealth/experimental offerings:
It's still up to me to figure out the best characteristics of each model to learn which are best for research, planning, implementation, etc. and maybe there's a future update in the works to help me find the right choice faster, but for now I'm excited to try them all out.
It turns your terminal into a true AI coworker pick whatever model you want and it becomes the place you generate, debug, and automate without leaving your flow. It’s for people who love the command line: fast, flexible, and powerful, with the model choice and control developers actually want.
Congrats! Are there plans to develop or integrate an automatic model recommendation system that can help users quickly identify the most suitable model based on their needs and context?
Having used opencode zen for a few weeks, I appreciate its curated selection of AI models tailored for developers.
I've been an opencode user for a little while now and so far have really enjoyed the zen experience over the last week. I haven't been diligent enough about subscribing to all the model providers out there to figure out which one I should be using for a given task, and zen has given me a single-entry-point solution to enable that sort of experimentation. I also appreciate that the models that are available in this list are coming recommended for coding with bespoke handling to squeeze the best juice out of each one.
The models available today include those listed below, plus a few stealth/experimental offerings:
@Claude by Anthropic
@GPT-5
@Qwen3
@Kimi K2
@Grok Code Fast 1
It's still up to me to figure out the best characteristics of each model to learn which are best for research, planning, implementation, etc. and maybe there's a future update in the works to help me find the right choice faster, but for now I'm excited to try them all out.