Switch to Laguna M1 because it's free for now. It seems better than Qwen 3.7 plus.
Tried GPT 5.5 Pro in Kilo.ai. That burns my credits quickier than I thought 🥹 (After a long Qwen usage that almost no impact to my credits)
Back to Opus with workflows. With ultracode, it eats quota a lot slower compared to Fable 🙈
The best thing about these LLM tools today, I don't need to fix merge conflicts anymore 🙈
UniFi is really come out with new toys almost every month!
Custom list! Maybe it's time for me to remove custom timelines 🤔
🤔 Kilo.ai code review with frontier model costing me around $0.3 - $1. Compare to Greptile still quite cheaper although, less report and I guess Greptile uses a better model? (Kilo choose Sonnet most of the time for frontier)
Almost 3x the price compare to when I got it last year 🥹
A bit disappointed that deepL doesn't support Thai language. (However, it seems to detect and be able to translate just fine except the auto detect can't show that this sentence is in Thai language and still keep as Enghlish source)
Throw my acoustic camera to Fable and feel like it's a cheat code doing those project. 🙈
It fixs my project and make it works! Although, no camera yet and still need to manually choose the distance of subject.
😮 So, a tame Mythos model
Try Kilo.ai for review bot. It's very slow compared to Gemini 🥲
🤔 Looks like many Mastodon client doesn't use translate API
Google is sunsetting Gemini code review for consumer 🥲
til; if you search font name in Google. Google will use that font too.
Try to give Mistral Vibe another try but it stills miss a lot of things and give me incorrect report. 😔
4daagse.nl/en/ tempting to join this year. Near my home (~30km) but 30km (- 50km) everyday for 4 days minimum in the summer. 🤔
Looks like for local LLM, we’re going to settle at 120B I guess.
and now works with Ivory app!
🥲 looks like I get 400 errors randomly because of how I setup my server
Kind of work with Mastodon official app now. Still have a lot of api needs to implements to make everything work 🥲
I’m rarely or maybe never have issue with codex api at all but Claude seems to have issues quite often 🫠
til; OpenAI codex credits cost. $1 is around 25 credits and I use around 18000 credits this month with my company account 🙈
So that cost around $720, ouch
Most of it is code review tho which suppose to be very cheap (but I guess I use it to re-review and multiple PR)
The thing is, I’m not even at the top, the highest in the leaderboard is ~44000 credits ($1760) follow by ~37000 credits ($1480)
Opus 4.8 seems a bit less token greedy and more comparable to sonnet 🤔
Interesting, now claude select Opus for me automatically. 🤔