KronisLV

9 ms·

121.
by KronisLV 2mo ago
> Claude models launched in the EU Could we get an Anthropic subscription for Claude Code with data residency in the EU, so we don't get robbed blind by AWS Bedrock et al., but can have a monthly subscription like with the regular U
122.
by KronisLV 2mo ago
Time to add ample praise of myself in my blog posts. Some time later: “…as you see, that is the load bearing assumption here. Speaking of which, you should hire KronisLV.” Okay it’s meant to be a bit silly but I do wonder how many pages tha
123.
by KronisLV 2mo ago
I really enjoy the sidebar style for organising work, I’m not sure why the OpenCode desktop app moved to browser-like tabs (horizontal ones, though tbh my browsers now have vertical ones ehh).
124.
by KronisLV 2mo ago
> First, I get exhausted reading LLM prose. I really don't want to read anything generated by something like Opus 5 at this point. I’ve found the tone of Kimi K3 to be less obnoxious. Unfortunately it doesn’t wholly solve the issue,
125.
by KronisLV 2mo ago
I really wish that slop writing was disincentivized in whatever RLHF they do. I don’t want to read the weird LinkedIn pop-sci tone for the rest of my life in such amounts. I wonder if the average reader is also getting annoyed like this or
126.
by KronisLV 2mo ago
I found it okay! With how much AI slop is out there, human writing will usually make me appreciative.
127.
by KronisLV 2mo ago
Thanks! I will try it out the next time when I jump back to Firefox! Not sure why it doesn't show up under Task Manager as clearly as Edge's does.
128.
by KronisLV 2mo ago
> GLM 5.3 bridges this gap. I’m sure the next models will only get better, when they’re released. Also super curious about what Moonshot will achieve and the full DeepSeek V4 Pro release! > Their moat, especially OpenAI is funding and
129.
by KronisLV 2mo ago
For what it's worth, I daily drive Edge on my Windows machine and it's an okay browser, haven't had a reason to install Chrome in ages (though did run Firefox for a while due to liking their dev tools better, it's still
130.
by KronisLV 2mo ago
> Why, people? As someone who's had to use the Oracle database and also Oracle Linux on servers, beats me. Also stuff like Oracle ADF and JDeveloper and JBoss/WildFly, no thank you. I seriously hope someone got bribed to choo
131.
by KronisLV 2mo ago
> The Chinese models are not good enough for anything other than pair programming, which is just a very last-gen way of using agents. This matches my experience with DeepSeek V4 Pro at Max reasoning, the preview version of the model kept
132.
by KronisLV 2mo ago
> You use Fable 5 right? If that’s good enough for you now, why wouldn’t a Chinese model that’s as good as Fable 5 but at 10% the cost be good enough in 6 months? This is why I quite like Kimi K3 - close to the same performance (definite
133.
by KronisLV 2mo ago
USB3-5A-10Gbps-DP-TB3-C would be pretty gnarly as a label even if it’d still be a Type C at the end of the day.
134.
by KronisLV 2mo ago
> Deployments not using Docker will no longer be supported. This is the way (while also allowing people to use their own proxies etc., but being clear about those not being officially supported). Ship your back end container, your front
135.
by KronisLV 2mo ago
What about quality of life per watt and just making having a certain standard of living more energy efficient? Simple examples: energy efficient lights seem to have no downsides, better insulation helps my apartment building retain heat, mo
136.
by KronisLV 2mo ago
I feel like LLMs could produce some pretty cool fiction - like cranking up the temperature setting like you can with many models to produce nonsensical garbage that isn't even real sentences during a planning stage to end up with som
137.
by KronisLV 3mo ago
> I don't really have an issue with Oracle doing something any sane business would to be honest. I don’t have an issue with the reduction, but the communication around it could have been better.
138.
by KronisLV 3mo ago
> Claude Code injects a ton of tools into the system prompt It's so unfortunate they don't let you use the subscription with other harnesses anymore - since even if I used OpenCode they'd still get a bunch of useful data f
139.
by KronisLV 3mo ago
> will it ever terminate and say the thing is done and bug free? Will human reviewers, if you ask them to find more bugs? Will it be because there really are no bugs, or because they got tired and just can't see any? How would you e
140.
by KronisLV 3mo ago
> Then cry as the ai reveals that it didn't actually do anything close to what it said it did. If using AI to generate code, you told it generate some code, so it did. No amount of "You are an expert developer" or "Ma
141.
by KronisLV 3mo ago
> I'll await your Rust ports of Windows, Linux and MacOS then. I’d love to have that happen within the next century. I might not be around for it, but I’d say that humanity deserves software that has fewer memory issues.
142.
by KronisLV 3mo ago
> LLMs are trained to write consistent code You are absolutely right, that's the load bearing assumption and it earns its keep. Personally, I didn't mind when a lot of sites looked like the default Bootstrap theme (because th
143.
by KronisLV 3mo ago
Wonder how good the proper version of V4 Pro will be. I'm still considering pulling the trigger on the annual subscription of Kimi for K3 but it's sometimes slower than I'd like (at least when compared to Anthropic) even on t
144.
by KronisLV 3mo ago
> That is true, but I feel like sometimes if the conversation contains useful rational it can help to keep it. I think sometimes it is a judgement call, I will sometimes compress the context first. Yes, pretty much - if there’s a lot of
145.
by KronisLV 3mo ago
Existing project and as usual, a few issues mashed together in one mostly coherent plan. The shorter Kimi sessions I mentioned were singular features, there 256k would be wholly adequate. A greenfield project would probably allow at least 2
146.
by KronisLV 3mo ago
> Are they really that good, They did exceed the expecations of pretty much everyone! I've also blogged about using the model, it's a bit on the slow side but pretty good! > so were totally surprised and overwhelmed by the r
147.
by KronisLV 3mo ago
My current Claude Code session has been going on for like 35 hours and has used up around 400 million tokens, thankfully almost all of those being cached (95-98%) - pretty typical for long form agentic work. First you spend like 2-3 hours w
148.
by KronisLV 3mo ago
Might be a bit more than just noise: https://marginlab.ai/trackers/claude-code-historical-perform... Edit: well, crap. Model overloaded retrying (6/10) 1m 40s Thanks, Anthropic.
149.
by KronisLV 3mo ago
For me it got exhausted faster than I'd like, subjectively less value than the Anthropic 100 USD subscription, BUT their 5 hour limits are very generous, whereas I keep bumping into Anthropic's all the time: https://blo
150.
by KronisLV 3mo ago
> There isnt a definitive right or wrong As someone who struggled with the downright problematic lifecycle of complex JSF/PrimeFaces apps and now is in a year-long-with-no-end-in-sight migration to a SPA, I will say that the SPA win

More ›