ExdayAI

one man | one dgx spark | one bilion dollar idea
ios & android app coming soon
#GoBigOrGoHome
#startup
#localai
DM to collaborate

[ExdayAI @exdayai]

@tysn_dev What about Qwen3.8 27b for single dgx?

[ExdayAI @exdayai]

@sfxnz Crying with only one😭

[ExdayAI @exdayai]

@sgkelrnvaroa Bro don't stack them on top of each other, that's too much heat

[ExdayAI @exdayai]

@gabu3d_pl i like SGLang more

[ExdayAI @exdayai]

@u1tra_instinct will it work on M4 pro, 48gb ram?

[ExdayAI @exdayai]

@MiaAI_lab I did expect around 48 but this is crazy

[ExdayAI @exdayai]

@prajdabre Now qwen3.8 27B it its dens model ai it will be slower than MoE but is's worth it in my opinion

[ExdayAI @exdayai]

@plKamilPawlik LM Studio na początek to chyba najlepszy wybór do nauki.

[ExdayAI @exdayai]

@LachTomek @wwydmanski @plKamilPawlik https://t.co/MiDelzNAcG
33 tokeny na dgx

[Mia @MiaAI_lab]

Do you like more speed? I know I do ⚡️

Run Qwen3.8-27B on your @NVIDIAAI DGX Spark with @sgl_project ✨

SGLang • up to 1M context • MTP

~33-35 tok/s single stream (structural)
~195-210 tok/s with 10 concurrent streams

No difference in speed when running with 256k or 1M context. Default is 256k to support 10 concurrencies with full kv cache.

Get it here:
https://t.co/rooS9ykL91

[ExdayAI @exdayai]

@divaagurlxw

[ExdayAI @exdayai]

@u1tra_instinct You are boss sir🫡

[ExdayAI @exdayai]

@MiaAI_lab deepseek vs ling what's the best for 1 dgx spark ?

[ExdayAI @exdayai]

@Tech2Wild We are fighting the bad guys 😂😂

[ExdayAI @exdayai]

@QwenDevs @ddalcu Shitttttt

[ExdayAI @exdayai]

@gabu3d_pl what is the best ltx2.3 vs wan2.2 vs minimax h3?

[ExdayAI @exdayai]

@ItsmeAjayKV @Alibaba_Qwen i love your content i do have dgx spark and i also use Qwen3.6 27B for me is the best one

[ExdayAI @exdayai]

@Tech2Wild I have no friends that's why i have to pick those 3 musketeers🙏
@elonmusk @JeffBezos @realDonaldTrump

[ExdayAI @exdayai]

@Tech2Wild i'm waiting🙏

[ExdayAI @exdayai]

@MiaAI_lab @NVIDIAAI I love your content