LIJaByXa

[LIJaByXa @LIJaByXa]

@SPA3TANx @pixelgirl @ElevenLabsDevs Drama 3

[LIJaByXa @LIJaByXa]

@teortaxesTex Ну не очень корректно, минимакс просто откалибровали модель быть уверенее из за этого сильнее галлюцинации и больше правильных ответов.

[LIJaByXa @LIJaByXa]

@canva @DesignArena Weights? Pls

[LIJaByXa @LIJaByXa]

@nanbeige @OpenAI @rasbt Hey guys any news about nanbeige 4.5?

[LIJaByXa @LIJaByXa]

@facedevstuff I think haiku is like 50ba3b, sonnet
500ba15b, opus 1.5-2t-a80b and fable/
mythos 4-5t

[LIJaByXa @LIJaByXa]

@MaxForAI @Zai_org Level of which fable? 5 or 5.1 of future 5.5?

[LIJaByXa @LIJaByXa]

@OpenBMB @aiwithsally Will u add support for miniCPM V4.7 35B on android apk?

[LIJaByXa @LIJaByXa]

@ChrisGPT Any info about glm 5.5 though?

[LIJaByXa @LIJaByXa]

@ChrisGPT Sir, any info about glm 5.5, qwen 4?

[LIJaByXa @LIJaByXa]

@Xianbao_QIAN I miss qwen 4 models, glm 5.5 models, kimi k3.1

[LIJaByXa @LIJaByXa]

@AndyDoruison @SophiaCupCup @YxzRainy I bet 10$ that the next deepseek pro will be <4t parameters

[LIJaByXa @LIJaByXa]

@SophiaCupCup @YxzRainy These rumors are about v5, not v4.1 pro

[LIJaByXa @LIJaByXa]

@PhotogenicWeekE https://t.co/FdFWkhFN3o

[vicc @viccpoes]

dw, there will be open-source greatness.

[LIJaByXa @LIJaByXa]

Pls make it identical to the model you host

[vicc @viccpoes]

dw, there will be open-source greatness.

[LIJaByXa @LIJaByXa]

@Xianbao_QIAN Pls say, that these models: nvidia, flux, muse werent the only one coming

[LIJaByXa @LIJaByXa]

@krishnavadithy1 @43fr192982 @aisearchio Not that guy, but i have really bad results with russian text

[LIJaByXa @LIJaByXa]

@bfl_ai Thank u guys

[LIJaByXa @LIJaByXa]

@bfl_ai Finally!

[LIJaByXa @LIJaByXa]

5/5

[LIJaByXa @LIJaByXa]

Bunch of results with krea 3

Pretty impressive

1/5

[LIJaByXa @LIJaByXa]

4/5