æm

🌫️

[æm @aemxm]

@JadenJTW @Actuallykeltan Assumption is that proprietary software companies will be stagnant

[æm @aemxm]

@Macro_Harder He knows the meme

[æm @aemxm]

Future high beings probably will love trump for naming them SI

[æm @aemxm]

@NousResearch @Plinz Mobile app would do the trick

[æm @aemxm]

@chasmmmmmmmmmmm He looks like zuckerberg from back

[æm @aemxm]

@dberenzon Wen tge btw?

[æm @aemxm]

@SpaceX @elonmusk Look at that..

[æm @aemxm]

I mean.. 😂

[æm @aemxm]

@alightinastorm Its on a flywheel…

more people using it -> more training data for ai -> ai gets better at using it.

New frameworks have to be really superior to surpass the older + its training data advntgs

[æm @aemxm]

@Priyannkaaaa With 2T params it will be cheaper than opus 5.5 but will it be better ?

[æm @aemxm]

@thdxr /queue is really handy

[æm @aemxm]

@Salgueiroficial Fuckin scammer

[æm @aemxm]

The biggest thing happened lately is @deepseek_ai 4.1 flash

It costs nothing and capable enough to do a lot of things

I haven’t seen anybody pushing it and observe the capabilities

How you achieve this by that much params!

[æm @aemxm]

@trikcode You can do order of magnitude more now with chinese flash models and it costs less than 20

[æm @aemxm]

@sattyyouneed Real question is do you do it to be polite or you want it know you are good on a hypothetical AI take over

[æm @aemxm]

@Teknium @micheltamanda Whats the total context token of hermes at default right now?
Last i checked was around 16k

[æm @aemxm]

@prasenx Yes

[æm @aemxm]

Magical things should happen when astra level models becomes 2x cheaper than junior level engineers on a pro rata productivity basis

[æm @aemxm]

@airport_girl Coolest machine evrr

[æm @aemxm]

@hooeem 8/10 Guinness line

[æm @aemxm]

@Polymarket Now remove the sound