Church parabola

Tech Enthusiast. 26. psionic enjoyer (iykyk). I post random things. dms are open. professional reply guy. check out my strawpage

[Church parabola @cseryphim]

@HocTap_SFM @givethegoat The drone and missile production capabilities argument don't hold water because I'm not aware of any African country that has the technology to produce the electronics and sensors needed for these missiles, drones and aircraft. They likely could produce the mechanical engines tho

[Nikita Bier @nikitabier]

The only group more insufferable than tech people is anti-tech bluesky users, who have been coming back to X in droves the last few weeks.

And they’re just discovering that the entire world got rich over the last two years while they were busy arguing with each other about politics and AI.

And they’re very, very angry.

[Church parabola @cseryphim]

@TokenGremlin I'm sure they will have great model soon but will they have the compute to serve them? and will my usage be enough to matter

[Church parabola @cseryphim]

@hunterxlane92 @buildwithrajath It's not just 3D but even coding/science benchmarks, Opus 5.5 is on par or beating Astra right now. Combined with the much more generous usage and tokens/second. It's a good time to be a Claude user but we all know the generosity won't last long after the IPO lol

[Church parabola @cseryphim]

@hunterxlane92 @buildwithrajath Me personally I have downgraded but I'm still on pro the rest of the month. The past few weeks have been very disappointing. Claude feels virtually unlimited right now. Hope OAI they find their way again

[Church parabola @cseryphim]

@TokenGremlin I'm so busy using Opus that I am forgetting about my codex usage... I'll just throw some random stuff at it to burn the tokens

[Church parabola @cseryphim]

@Ananth7e Yeah I mean... I'm really happy with Claude right now. 5.5 are all insane. 6.1 Sol is good but it's just not good enough. I am switching to the $20 plan next month and token maxing some projects to get rid of my tokens but I don't think I'll be going back to a pro plan soon

[Church parabola @cseryphim]

@TokenGremlin I think its likely it will be better at Code but will it have the same kind of taste and ability to make 3D models and such as Fable 5.5? imo this is a bigger moat than coding benchmarks. I've been using Opus 5.5 for game dev kind of stuff and its far and away better

[Church parabola @cseryphim]

@Ananth7e Keep this up and you can maybe almost afford to get the $500 subscription and drain your week of usage in 30 minutes using ultrafast!

but for real, congratulations dude. Love your takes and the humor

[Church parabola @cseryphim]

@TokenGremlin @meta_x_ai Big thing, never really refining their products outside of chat. Codex is the exception but we had the original codex, canvas, they've relaunched plugins like 3 times, etc. anthropic is very focused on just doing smart models and agents which helps them a lot

[Church parabola @cseryphim]

@TokenGremlin @meta_x_ai I also generally prefer OAI because I feel they offer a more comprehensive product. Voice and image gen are underrated and there is nothing like ChatGPT.

However I think they are very unfocused and have too much going on. It feels like they are constantly sprinting to the next

[Luke @Luke_The_Tako]

i need yall to look at the size of these puppies
aight bye

[Julian Schiavo @_julianschiavo]

based on my data analysis, integer multiplication will be linear time by tomorrow!

[Church parabola @cseryphim]

@TokenGremlin You keep gassing it up hardcore and I'm just waiting for it to be available on my phone since I'm away from the computer all day

[Church parabola @cseryphim]

@TokenGremlin @thsottiaux But unironically it's annoying how puritanical these chat apps are. Like sometimes violence in creative writing or descriptions of guns can trip them. Or even image gen requests that are pg13. Extremely annoying. Kids shouldn't even be on this app, that's what teen mode is for

[Church parabola @cseryphim]

@TokenGremlin This is great but I wonder about token efficiency. Token efficiency is very important for these models imo because they're the ones I actually would use the API for. If I'm buying a subscription I'm not using Luna or Haiku

[sage @sagespells]

my favorite kind of intimacy is when two people start using each other’s random phrases and mannerisms without even realizing it. you just wake up one day and realize you speak a little bit of their language.

[Church parabola @cseryphim]

@TokenGremlin And for clarity, rn codex doesn't have ur chatgpt memories and vice versa. This makes it really annoying to work between them sometimes because it doesnt have all that context about me

[Church parabola @cseryphim]

@TokenGremlin Merge memories and context but Chat is literally the main product and the most used thing for me. Rn the big things keeping me on ChatGPT subscription (i reduced from pro to plus) is image gen and chat and the memories/context it has