> On the other hand Luna 6 on 20$ subscription from OpenAI cover by usage for whole month. However, `ccusage codex` command suggested that my Luna6 token cost is around ~100$ monthly.

This is where the model comparisons become hard. Luna and DS4.1-flash aren't comparable models in terms of coding performance, DS is far ahead. Or at least so say the benchmarks. So if you're good with the performance from Luna, there isn't even really a reason to look at DS. You'd pay more for capabilities you don't need. But that doesn't answer your main point of, what to do if you DO needs DS4.1-flash performance.

> And what I'm wondering is, as suggestted by very cheap cost-per-token from DeepInfer, why don't Deepseek persued subscription market as well ? What make subscription not make sense for them to provide ?

By bet, given that they've introduced time-of-use pricing, is hardware capacity. They don't have the capacity to handle more demand, which is what coding plan would immediately produce.