OpenCode Go has adjusted its model quotas. According to publicly available information, on the $10-a-month plan, DeepSeek Flash's monthly quota has climbed back up from a previous $15 to $30. The Pro tier stays at $15, and the service is starting to introduce peak and off-peak pricing.
The change isn't just about "more quota." Plans built for AI coding are usually bought for continuous workflows: a repository search, a multi-file edit, and a few rounds of testing — usage that is anything but even. Restoring the quota gives lightweight models more room to run, while peak and off-peak billing puts the cost signal back into the time of use.
Two figures worth separating
The $30 and $15 here are model-usage quotas built into the plan — they are not the same as a user's actual bill, and they can't be directly compared to per-token API prices from other services. Exactly how the peak and off-peak rates are calculated, which models they apply to, and whether unused balance carries over between billing cycles all still depend on the platform's final billing rules.
For developers, it's worth watching whether peak hours make it easier to hit rate limits, and whether the restored Flash quota is enough to cover everyday searches, autocompletion, and small edits. Once a price sheet turns into a set of dynamic rules, a clear breakdown of usage matters more than one eye-catching plan number.
Sources: OpenCode service information, CocoLoop; verified Go plan pricing, the Flash and Pro quotas, and the peak/off-peak rate adjustment.