merd

hello.

[merd @merdified]

@Snoopy_Osc Oscar P4astri 😭😭

[merd @merdified]

@RinoTheBouncer yo this guy pulled out a dead man's switch for a game leak

[merd @merdified]

@laurogeitabat LET'S GOOO OSCAAAAR jek PIASTRIII

[merd @merdified]

@KrystalWolfyAlt chat, bot, pinned (basically a bulletin board of funny quotes) and that's about it. 👍👍👍

[merd @merdified]

@EliasAMurphy I'd recommend https://t.co/FuGKkNbdUL and/or https://t.co/9qfDvGRwN7

[merd @merdified]

@Vijay88178864 @ryanvogel I guess they're using the Monkai theme

[merd @merdified]

@opencode OPENNCODEEE RESET MY FREE USAGE AND MY LIFE IS YOURSS

[merd @merdified]

deepseek had enough with my bs and closed my pc for no reason i'm crine 😭😭😭

[merd @merdified]

@f4micom reasonable reaction.

[Lvya @Lvya13]

君にまつわるミステリー

#氷菓

[merd @merdified]

@f4micom what do you mean? that's just regular ol' theo yk

[merd @merdified]

@mattlam_ Very interesting! According to the benchmark, pi seems to be the best choice for deepseek-v4-flash. I really like using opencode but I'll give pi a try only because of this!

[merd @merdified]

@thdxr fair 👍🏻👍🏻

[merd @merdified]

@ErickSky I've tried it and I don't think it's very accurate in STT. However, that might be because I am not a native English speaker, so my accent might be a little off.

Considering the size of the toolkit, cool project.

[merd @merdified]

@EntelligenceAI Opus 4.8-level performance for THAT price point is hard to believe. If they somehow pull that off, uhh, respect.

[merd @merdified]

@stevibe I wonder if relatively small yet okay models like Qwen3.6 35B A3B can be even smaller with 1-bit quantization