bel8

5 ms·

61.
by bel8 2mo ago
Funny, I just ran the test myself and DS4 got it right first try. Then I ran it 9 more times and it got right 9 out of 10 times. Lesson learned to double check what I read on the internet.
62.
by bel8 2mo ago
Well it succeeded on first try for me: https://i.imgur.com/gljOYr9.png [Image 1] what time does the clock show? + Thought: 368ms The clock shows 8:10. - Blue hour hand: just past 8 - Gre
63.
by bel8 2mo ago
28 crates for 3 of the most common functionalities. And it would have been 0 crates in .NET or Go. Even after adding a web server.
64.
by bel8 2mo ago
I disagree. It's not even that useful to train LLMs to read an ancient analog clock. Unless we're talking about AGI, I couldn't care less if an LLM is bad at things they won't be doing anyway. I'd rather focus train
65.
by bel8 2mo ago
It still lacks basic functionality like a JSON parser, regex, directory walker, rnd generator and cli arg parsing. I won't mention lack of date/time lib because that's complex and changes often. That's why projects end u
66.
by bel8 2mo ago
So it's a Fable class LLM? DSV4Pro vs Fable5 HLE w tools 60.0 vs 63.0 Terminal Bench 2.1 87.9 vs 88.0 Cybergym 83.3 vs 83.1 DeepSWE 62.7
67.
by bel8 2mo ago
It's trendy to hate on Meta just like it's trendy to handwave on Apple. One can do no right regardless, the other can do no wrong. At least in HN.
68.
by bel8 2mo ago
From opencode go $10/mo plan I get between 60 t/s and 100 token/s even with large contexts of 150k+ tokens. I wouldn't call 80 t/s slow.
69.
by bel8 2mo ago
I've been using for work, from opencode $10/mo subscription plan, on high effort (which is better than max imo), and haven't had any issue. When it was first available in opencode, it was kinda slow for me, I guess because ev
70.
by bel8 2mo ago
It was. People are overreacting.
71.
by bel8 2mo ago
So whoever is using DeepSeek V4 flash gets a massive upgrade without having to change any config.
72.
by bel8 2mo ago
Yes I will tyvm. They release the models back for free.
73.
by bel8 2mo ago
flash for code: c#, TypeScript, PHP, SQL, CSS, HTML. also features, tests, fixes, refactoring and planning. it's super fast, smart and dirt cheap.
74.
by bel8 2mo ago
I have the impression that SolidQueue uses whatever your rails app is using by default. That can be SQLite, PostgreSQL, MySQL, etc. So I don't think they changed to PostgreSQL per se.
75.
by bel8 2mo ago
Yeah that's why I asked. It would be one thing to use, another to host. And it seems nobody has a source so it's just humors as usual.
76.
by bel8 2mo ago
source? edit: it seems asking for a source it frowned uppon in this site. And it seems there's no source.
77.
by bel8 3mo ago
Quasar is amazing. But I dislike their Material design components. Would be perfect otherwise.
78.
by bel8 3mo ago
Perhaps they started banning entire categories of projects so their infra could actually work again. Banning a portion of their community is... one way to reduce server load.
79.
by bel8 3mo ago
Curious about what in the article convinced you.
80.
by bel8 3mo ago
I moved my domains from GoDaddy to NameCheap some years ago. And recently from NameCheap to Cloudflare once I heard NameCheap changed owners. So far, so good. If Cloudflare messes up my domains, of all things, I might as well quit tech and
81.
by bel8 3mo ago
could you share the prompt? that looks useful!
82.
by bel8 3mo ago
They don't have to be affected to be worried. It's how it was done and possibilities of further bans that is worrisome. I have small open source projects that I won't move to Codeberg ever now because I don't want to was
83.
by bel8 3mo ago
I too love and use pi agent. But all of these agents have or had "dumb" security issues: https://github.com/anthropics/claude-code/security/advisorie... You probably know that since you run pi insid
84.
by bel8 3mo ago
The RCE they mention was fixed in December, 2025: https://github.com/anomalyco/opencode/releases?page=21#relea...
85.
by bel8 3mo ago
I would agree if openrouter was the only provider of those. But those LLMs are also offered outside openrouter.
86.
by bel8 3mo ago
https://openrouter.ai/rankings#top-models Chinese models dwarf USA models usage. And now there's a Fable/5.6 alternative. The gap widens. Now go and ask for your GP poster for their data as well. Unless you'r
87.
by bel8 3mo ago
Continued how? I have switched most of my personal LLM coding to DeepSeek V4 Flash since it was launched. And now 100% to a mix of K3 / DeepSeek V4 / MiMo 2.5. It's nice not being called a terrorist just because I told it to
88.
by bel8 3mo ago
That's quite small for my workloads. I try to keep it under 200k but my DeepSeek and MiMo sessions can sometimes grow to 350k tokens when I try to squeeze one last iteration I compact. Can't OpenAI copy DeepSeek K/V cache tec
89.
by bel8 3mo ago
Contrary to many, I have high hopes that Jarred and team will lapidate the ported codebase and it will continue to flourish as an open source project.
90.
by bel8 3mo ago
Cool, and how is it ranked?

More ›