gpt 6 astra for excel when? @thsottiaux

man astra is so good on light especially computer use but dang those limits are draining 😖

anyone’s else’s grok 4.6 in @cursor_ai super duper slow, making me have to go back to 5.6 sol

my 5x openai sub just got absolutely destroyed and idk if it’s placebo from all these tweets about useage 😓

Qwen3.8-27B went from 26 tok/s to 62 tok/s this is what decentralization looks like

people are finally noticing the rate limits on closed ai subscriptions keep getting tighter it was always coming compute costs keep rising and those plans were never priced to the real cost of inference meanwhile qwen3.8-27b just dropped fully open weight runs on consumer hardware and already outperforms a lot of the closed options people are still paying for such as luna local ai is the only real fix :P

If astra is smart enough, do you guys think OpenAI will further train Luna in another model update 🤔