VolatilityX_intern

Welcome to this VolatilityX Update! Announcement: @XVolatility now supports crypto news and analysis. $Voltx

@orangie Minimax 2.7 Pretty good! But not opus

what models are you using now for your openclaw since the Claude ban

@kenwheeler minimax 2.7

what’s the best non anthropic model to run claw on?

Let's go! Peace time hopefully

wtf just happened

Nopes, they are testing our limit of recognizing AI and train on that!

1 decent bug and startup files for bankruptcy!

@alexanderrX_ I think can guess the building looking at your profile pic

i am of course very bias as it’s also my residence

And it did! Went all the way to the ground! Wow!

@milalolli Hi from the building in the pic

Londonmaxxing to the max! Views from Canary Wharf - the Wardian

@katiabanina I think the Dubai crowd has arrived :)

how is Londonmaxxing suddenly all over the timeline

@Yuchenj_UW @karpathy my guess was it spins off subajents and they fail for some reason but the main agent doesnt know (happens in Opus 4.6 as well) but yours is totally different.

GPT-5.4 xhigh seems bad at following instructions. Last night I launched two AI research agents running @karpathy’s autoresearch. Claude Opus 4.6 (high): > ran for 12+ hours, 118 experiments done, still running GPT-5.4 xhigh: > stopped after 6 experiments > blamed me for “manually interrupting” it > I interrogated it > It admitted it made a mistake and stopped the loop itself, despite an explicit LOOP FOREVER instruction in the md file. 💀

Yes! this is the most frustrating thing! it gets worse with OpenClaw.

All numbers are inflated by 100% at least!

@kr0der mostly medium, switch to high if it gets complex. 75-80% medium

what level of reasoning are you using and finding success with for GPT 5.4? are we still spamming High or is Medium the new meta?

@Zeneca have been able to handle the dumb head thing. but I am seeing a lot of time the main agent takes a task but doesnt do it ... if i recheck it says taking time. pretty odd! checked the logs as well. I think subagent spawn issues but thats frustating

To the people who say openclaw is absolute magic and increasing their productivity 100x... Are you still having to deal with it being a big dumb dumb buggy head several times a day? Where things just break randomly? Or do you truly have a magic system and setup where everything is just glorious and works perfectly?

If the coding session goes for a few hours does GPT and Claude both degrade in service or it is just the mistakes that start to surface?

Contributing to londonmaxxing

Londonmaxxing that was a surprise! If we can get the DeFi summer vibe again would be awesome!

@the_smart_ape @AnthropicAI Am seeing sudden jumps in usage in my max 200 plan. For same usage now I consistently hit my weekly limits

lately i've been hitting my Claude token limits way faster than before. same workflows. same complexity. nothing changed on my end. so i started digging. and what i found explains a lot. @AnthropicAI is bleeding money on Claude Code. a $200/month Claude Code subscription was burning ~$2,000 in compute costs last year. today? that same $200 plan consumes up to $5,000 in compute per month. that's a $4,800 loss. per user. per month. but this is a classic playbook, burn billions to capture the market. subsidize everything. make the product feel cheap. then slowly tighten it. we've seen this before with @Uber where rides used to cost $5. DoorDash delivered for free. and it's not only Anthropic, all these AI tools, Claude Code, Cursor, Copilot, Codex, are priced below cost right now. every single one. the entire market is subsidized. when the subsidies stop (and they will), one of two things happens: 1. prices go up significantly 2. features get throttled or both. probably both. this is why i think my token limits feel tighter. Anthropic is slowly adjusting in a quiet, gradual, "you probably won't notice" way. less compute per dollar. slightly more aggressive rate limiting. i'm not mad about it. it makes business sense. you can't lose $4,800 per power user forever. but if you're building on top of these APIs, and i am, you need to plan for a world where AI compute costs 5-10x what you're paying today. the cheap AI era will end. build accordingly.