Beda

[Beda @beda098281867]

@HotAisle I assume mostly Grok 4.6 and Fable
and one thing I noticed, Opus 5 does a much better job in Cursor then in Claude Code, it stays better on track for research stuff and the code it writes is somehow less sloppy too

[Beda @beda098281867]

@mntruell bring them back please! OpenAI models are quite useful and to some degree work better in Cursor then in the Codex harness. Figure out a way to resolve this please!

[Beda @beda098281867]

@Sentdex finally time for "GLM plays GTA"

[Beda @beda098281867]

@TheAhmadOsman @sama totally agree. refusals without trusted access are getting out of hand
almost as bad as Claude at this point.

5.4 was fine, 5.5 got worse, 5.6 Sol refuses anything that even smells like security
Daybreak is the only thing that actually works, but it eats token quota alive

[Beda @beda098281867]

@thsottiaux sir you really need longer hair and longer beard

[Beda @beda098281867]

@ZixuanLi_ did you guys get more compute? pretty impressive what you pulled off with all that free ox alpha serving. is the coding plan now powered by the same machines?

[Beda @beda098281867]

@sama throw simultaneous launch parties at every OpenAI office worldwide

[Beda @beda098281867]

@theo paying for Fable and getting Opus 4.8 instead. Refusals are exhausting and keep getting worse. if 5.1 is even more trigger-happy, what’s the point?

[Beda @beda098281867]

@leerob @bot

[Beda @beda098281867]

@bot

[Beda @beda098281867]

@rxmphai @thsottiaux @s_batzoglou or maybe to celebrate the announcement about their new Jalapeño chip

[Beda @beda098281867]

@LuxuryWatchGuy1 @sama nothing beats the F-91W, do you customize those too? ;)

[Beda @beda098281867]

@thsottiaux @s_batzoglou what was the "not so random" reason? something to celebrate? did you guys reach 21M or something? now I'm curious :)

[Beda @beda098281867]

@thsottiaux sir you need longer hair and longer beard, that would look impressive!

[Beda @beda098281867]

@thsottiaux @ziwenxu_ let the hair grow man, long hair Tibo + beard would surely look sick

[Beda @beda098281867]

@leerob any news about official Grok Bot for Linux?
I know, there are community builds but official support would be great

[Beda @beda098281867]

@MiaAI_lab don't forget us Linux AMD users :)

[Beda @beda098281867]

@thsottiaux thanks for the transparency Tibo!
Pro without the 5h limit is great to hear, hope you guys will keep that forever. No 5h limit allows me to work way better!

[Beda @beda098281867]

@MiaAI_lab a setup with good quality, decent context and still usable speed that could fit in an 7900 XTX 24GB would be great

[Beda @beda098281867]

@0xSero don't forget older AMD cards like 7900 XTX