Artem

Working on AOSP-related things at @ActinisHQ.
Leading the apps team at @GrapheneOS.
Opinions are my own.

[Artem @integer_32]

@IndependentEco I guess they use the same agency that asked influencers to post a Chinese foldable review on the foldable iPhone release day 😀

[Artem @integer_32]

Kimi K3.1 Pro is the funniest scam ever. Bought a $200 sub to test; hit the 5h limit in 20 min. After that instructed the agent to NOT use subagents (why it's spawning unlimited subagents by default??) - 5h is still sad, but at least doesn't die in 20 min. Cancelled sub; will never buy again.

[Artem @integer_32]

Even GPT Astra's limits are not that bad currently, lol.

[Artem @integer_32]

@suchenzang It's in fact much more expensive than even Astra, at least on their plans.

[Artem @integer_32]

@_MaxBlade Yep. Same shit, even current OAI's limits are much better.

[Artem @integer_32]

@Hesamation K3.1 limits are a joke (and their harness seems to be quite bad).

[Artem @integer_32]

@Kimi_Moonshot Ran a code review task on the $199 plan with default settings; it consumed the 5h limit in 20 minutes and stopped, lol (default speed). Perhaps it used too many subagents, but it's still a problem: I didn't ask it to do that.

[Artem @integer_32]

@Link4Christ @GrapheneOS You have to partner with every carrier (even Apple doesn't do this), or to use Google's monopolized services, which are exclusive to Google's Messages.

[Artem @integer_32]

@thekitze [Object object]

[Artem @integer_32]

@veryayskiy @Revolut Nothing new, shitty "bank" acts like a shitty bank.

[Artem @integer_32]

@OpenAIDevs Nothing - you can't with such limits. $200 feels like $20.

[Artem @integer_32]

@L_33_H A non-existent car can even do 0.05s without a gas thruster 😀

[Artem @integer_32]

@LSK94LH @TheMindScourge 13k car competes 35k+ car? Sure? Why not a Rolls-Royce?

[Artem @integer_32]

@TheMindScourge Also, the second one exist. The first one not yet, and not sure if it will.

[Artem @integer_32]

@CodexResets1 Same on $200 plan, it's almost realtime

[Artem @integer_32]

@maelguimet @thsottiaux Then why does $200 of Opus 5.5 last for a week on the same project with the same tasks?

[Artem @integer_32]

@ValueRaider @Vincent20081825 @thsottiaux Not sure that it won't last. Inference itself is cheap (but yep, there are other costs like training), so I would say that the API is overpriced rather than the subscription being subsidized.

[Artem @integer_32]

I have 2x $200 anthropic subs and 1 ,$200 OpenAI sub, and I'm really a heavy user of all of them. Earlier OAI felt unlimited and Anthropic was really limited. Then, after Opus 5.5 release Anthropic was really unlimited (it was impossible to hit 5h or 7d even on a very big project), and OAI became very limited at that point. Now Opus 5.5 is much more limited, but still eternity better that sol in terms of both limits and intelligence

[Artem @integer_32]

@Vincent20081825 @thsottiaux Anthropic's $200 sub gives much more weekly usage comparing O5.5 vs Sol6.1 (not taking into account that Opus 5.5 is much smarter)

[Artem @integer_32]

@thsottiaux And I can provide a real proof of a not-that-big project, @thsottiaux

[Artem @integer_32]

@thsottiaux Usage is still really bad - you still can drain your 7d in 1 day