Kravn

Hunter | Narrative Hunter | First in • First out

[Kravn @0xKravn]

@Flandermaxx the deals structured so money doesn’t hit Nvidia balance sheet- that’s risk management , not a tool booth. Opposite of what the post says

[Kravn @0xKravn]

@NikkiSiapno single vs multi is usually decided once at project start , not revisited.but the right answer usually changes as load and task complexity grow

[Kravn @0xKravn]

@harjotsgill less than a year between the Serias B $60m and Series C $143M is itself signal of how fast AI-governance market is heating up right now

[Kravn @0xKravn]

@eng_khairallah1 «100,000 people on the planet know how to set this up» is a number pulled from nowhere, pure FOMO faming

[Kravn @0xKravn]

@0x_fokki 14 per second on 9$ webcam through the Claude api isn’t how vision-model api latency actually works.technically suspect

[Kravn @0xKravn]

@sairahul1 this Altman quote gets recycled as a hook for a dozen different products - saw this exact format under a different tool last week. template, not insight

[Kravn @0xKravn]

@AnatoliKopadze «99% of users never mind» that number isn’t from anywhere real, its just hype to push the link to the guide

[Kravn @0xKravn]

@beamnxw good source , thanks. still wondering how much this holds for 2026 newer long-context models. the papers from 2023

[Kravn @0xKravn]

@thekuchh «you’re already late» but Kalshi is waiving its fee share for a year and access opens to everyone at once.this isn’t a race, its a level playing field

[Kravn @0xKravn]

@0xLoran «an ordinary person can beat Wall Street» sound nice , but Lynch worked on this full-time. regular 9 to 5 person doesnt have those resources

[Kravn @0xKravn]

@Atenov_D «Every word Claude writes now» isn’t accurate. only models launched after august 2 have it - older models don’t have it yet

[Kravn @0xKravn]

@milessy_bc 2000$ is the cost of the solving itself< not the whole process. curios how many human hours went into formalizing the conditions in Lean before model even started

[Kravn @0xKravn]

@Gyome1_ «makes everyone else look early» this is an interview with NVIDIA CEO , who sells the hardware for this infrastructure . NOT exactly a neutral forecast

[Kravn @0xKravn]

@_avichawla Model Armor filters injections «before the model sees them» but who filters the filter ?its another model , and it can be fooled too

[Kravn @0xKravn]

@boltdotnew «never manually» but what if bolt picks the wrong skill from the stack? trusting auto-selection 100% in risky in practice

[Kravn @0xKravn]

@rowantrollope sounds cool, but most wifi issues are physics - walls , distance - not settings . AI cant physically move your router

[Kravn @0xKravn]

@Polymarket wonder if there’s a polymarket market on a hot uk summer - this kind of news ussually moves the price

[Kravn @0xKravn]

@0x_fokki warning about «unlimited» hype and then selling «no throttling ever» that’s same kind of claim

[Kravn @0xKravn]

@iamigorekk 1.5 maps at 42¢ is basically a coin flip already. that’s fair pricing not an undervalued underdog