What is free, and what do you pay for?
Official service details checked October 25, 2026. OpenCode’s older Zen documentation now points to Console model access. Codimixa is a separate API service, with its own account and billing.
| Access choice | Cost basis | What to check |
|---|---|---|
| OpenCode client | Free, open-source software | Model access and any paid tools remain separate. |
| OpenCode Console / Zen | Pay-as-you-go model requests | Current model rates, credit and billing terms. |
| OpenCode Go / Go Plus | Listed at $10 / $40 per month | Included models and usage limits; these are not unlimited plans. |
| Your Codimixa API key | Usage at the selected Codimixa model and account rates | Exact ID, token categories and account billing conditions. |
| Local models or free offers | Depends on your chosen setup | Hardware costs, eligibility and offer limits. |
Budget a coding session, not a single prompt.
A session can include repository context, tool results, follow-up edits and test failures. Measure a complete task from the first instruction to an accepted patch.
A worked API cost example.
Illustrative workload: 100 small coding tasks. Assume one API request per task with 20,000 uncached input tokens and 2,000 output tokens. Additional turns and retries add requests or tokens. These are budget assumptions, not client benchmark results.
Model: gpt-6-luna. Account group: default. Rates observed 2026-10-31; publisher references checked 2026-10-20. Context tiers are selected per request. The comparison uses standard publisher processing.
Cache reads and writes are zero in this example. Hosting, subscriptions and paid tools are additional. Change the token mix in the calculator to fit your own work.
| Budget input | Assumption or estimate |
|---|---|
| Model / exact Codimixa ID | gpt-6-luna |
| Requests or complete task samples | 100 |
| Uncached input / output per sample | 20,000 / 2,000 tokens |
| Codimixa input / output, USD per 1M | $0.10 / $0.50 |
| Estimated Codimixa API total | $0.30 |
| Estimated official API reference total | $0.30 |
Estimate from work you actually repeat.
Run a small set of typical tasks. Divide total API cost by accepted tasks, then multiply by your expected monthly task count. Keep small edits and larger refactors in separate groups if their costs differ.
| Task or recurring job | Runs per month | Measured API cost per completed run | Monthly estimate |
|---|---|---|---|
| Your representative task | Enter expected runs | Use measured API cost | Runs × cost per run |
See what a lower rate changes.
Use each model’s current Codimixa rates for the sample. Compare with matching official rates while keeping token counts the same. This shows the unit-price effect separately from changes in the amount of work.
Choose based on the work you expect to run.
For a subscription, compare the included models and limits with your expected sessions. For usage billing, start with a small task budget and inspect the actual charge. A flat monthly fee and a per-token estimate cover different conditions; neither establishes which setup costs less for your work.
A FEW DETAILS
Before you continue
Do I need Go to use OpenCode with Codimixa?
No. Configure Codimixa as a custom provider with a Codimixa API key. OpenCode Go is an optional service; its subscription and included usage do not pay for requests sent to Codimixa.
Does a model access plan cover a separate API key?
Check the terms of the plan you choose. Treat requests made with a Codimixa key as Codimixa API usage rather than assuming another plan includes them.
Will a cheaper model always reduce my coding bill?
A cheaper rate helps when the model can complete the task. Extra retries, larger outputs or manual repair can change the cost of accepted work.