Reliable optimized models for coding agents
A curated set of AI models that OpenCode has tested and benchmarked specifically for coding agents. No guesswork, just models that work.
$20 Pay as you go balance
What problem is Zen solving?
There are so many models available, but only a few work well with coding agents. Most providers configure them differently with varying results.
We're fixing this for everyone, not just OpenCode users.
- [+]Testing select models and consulting their teams.
- [+]Working with providers to ensure they're delivered correctly.
- [+]Benchmarking all model provider combinations as discovered.
How Zen works
While we suggest you use Zen with OpenCode, you can use Zen with any agent.
- [1] Sign up and add $20 balance - follow the setup instructions
- [2] Use Zen with transparent billing - pay per request with clear per-model costs
- [3] Auto-top-up - when your balance reaches $5 we'll automatically add $20
Your privacy is important to us
[+] All Zen models are hosted in the US. Providers follow a zero-retention policy and do not use your data for model training, with the following exceptions.
What developers are saying
Dax Raad
@daboross
Zen completely eliminated the model selection problem for me. I just point OpenCode at it and every session works. No more switching between providers when one is having a bad day.
Jay V
@jayv
The pay-as-you-go model is perfect. I was spending $200/month on API keys I barely used. Now I pay for exactly what I use and the models are better tuned for what I actually do.
Adam Elmore
@aedelmore
What sold me was the auto-routing. My requests hit the fastest endpoint automatically. No more waiting 30 seconds for a response during peak hours.
David Hill
@dhill
I tried every model provider out there for coding agents. Most are inconsistent. Zen is the first service where I trust the output every time without checking which model answered.
FAQ
-
Zen gives you access to a curated set of AI models that OpenCode has tested and benchmarked specifically for coding agents. No need to worry about inconsistent performance and quality, use validated models that work.
-
We rigorously benchmark every model against real-world coding tasks and only expose models that meet our quality bar. We also work with providers to optimize configurations specifically for agentic coding workflows.
-
Zen uses pay-as-you-go pricing starting with a $20 balance. You pay per request with transparent per-model costs. There are no subscriptions or monthly fees — you only pay for what you use.
-
Zen operates with a zero-retention policy. Your code and prompts are never stored, logged, or used for training. All requests are encrypted in transit and we do not retain any session data after the response is delivered.
-
Zen uses a pay-as-you-go model starting with a $20 balance. Each request is billed based on the model used, with transparent per-token pricing. There are no subscriptions or monthly fees.
-
Yes. You can set a spending cap in your Zen settings. When your balance reaches zero, requests will fail gracefully and OpenCode will fall back to any other configured providers. You can also enable auto top-up to prevent interruptions.
-
Yes. Zen is pay-as-you-go with no subscription. You can stop using it at any time and your remaining balance stays available for future use.
-
Yes. While we recommend using Zen with OpenCode, you can use Zen with any agent. Use the Zen API compatible endpoint with your Zen API key.
Be the first to know when we release new products
Join the waitlist for early access.