Tibo 28-day challenge day 20: Codex gains and a GPT-6.1 Astra rumor
The author lists improvements from the 28-day challenge, including faster models and persistent caching in Codex, and speculates that OpenAI resuming RL training could bring 6.1 Astra within ten days.

tibo 28-day challenge is going great it’s day 20 and we’ve already gotten 9 resets. today might be another one as every saturday so far has been a reset lmao we’ve gotten some really good improvements too > faster models > fewer "model at capacity" errors > persistent caching in codex, biggest one for me > more efficient context reuse > and a lot more too and now openai reportedly resumed RL training on 1st oct which makes me think we might get banger model in the next 10 days which could be none other then 6.1 astra, what do you think?