ltcabral

[ltcabral @ltcabral]

@thsottiaux Too late… I’m at 0% now

[ltcabral @ltcabral]

@thsottiaux Still didn’t get Astra on my Pro 20x desktop app

[ltcabral @ltcabral]

339 working. Bro accidentally founded a company.

[ltcabral @ltcabral]

I also tested Gemini 3.5 Flash-Lite, which performed worse than 3.1 in some of these use cases. That makes Luna’s result even more compelling. I’m genuinely excited about what comes next.

[ltcabral @ltcabral]

Luna’s new price may be one of OpenAI’s best recent moves.

It makes the model an exceptional option for processing large volumes of structured data and powering iterative systems that need low latency, not just a cheaper general purpose model.

[OpenAI @OpenAI]

We are committed to pushing the model frontier across cost efficiency, capability, and speed.

Starting today, we are reducing prices for GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20% , and offering a faster option for GPT-5.6 Sol in the API.

Luna and Terra’s lower prices are reflected in how usage is counted in Codex and ChatGPT Work, so your usage goes further.

[ltcabral @ltcabral]

Gemini 3.1 Flash-Lite was the first model I found that consistently outperformed it. I also tested several open weight models. Some were strong on individual checks, but none passed the full suite for instruction following, structured outputs, tool calls, routing, and latency.

[ltcabral @ltcabral]

@gdb @OpenAI Love that I can keep the company knowledge up to date by navigating through the product autonomously and reading Linear boards and everything else daily.