bel8

6 ms·

121.
by bel8 3mo ago
Which features?
122.
by bel8 3mo ago
SWEBench Pro should be ignored until they fix it or disprove the broken task accusations.
123.
by bel8 3mo ago
It seems to trade blows with GPT 5.5 and Opus 4.8 in performance while being cheaper than GLM 5.2.
124.
by bel8 3mo ago
> Claude Code v2.1.181 (released June 17th) and later use the Rust port of Bun. It seems the reports of Bun's death have been greatly exaggerated.
125.
by bel8 3mo ago
> he trigger was watching deepseek-flash fail on the simplest /review run, every shellCommand and readFile call bouncing back with a raw zod issues blob, the model unable to recover because the error wasn't in a form it could r
126.
by bel8 3mo ago
From what I understood, the gist of it is that uses a bunch of skills in a loop to help LLMs solve the tasks better via browser-testing, debugging, visual-testing, etc.. The skills can be found here: https://github.com/ironb
127.
by bel8 3mo ago
"Fable" is the new "Rust". If your HN title contains "Fable", you get instant upvotes just because, regardless of merit.
128.
by bel8 3mo ago
yeah, and might as well ban Chinese models too for safety since they can do the same.
129.
by bel8 3mo ago
This looks like the usual https://threejs.org animation built by LLMs. I'm more interested in: - The correctedness of the data and if it was reviewed. - The skills and challenges while building it. Was it a one shot? Did it
130.
by bel8 3mo ago
Siri fails simple tasks often, I don't know about less wrong output.
131.
by bel8 3mo ago
Kudos to you. Every time I try to vibe code myself to sleep I blink once and it's 5am. Creating is too much fun. Any tips?
132.
by bel8 3mo ago
Title is click bait. This started back in February and looking at commits, Fable did only a small part of the latest commits. 19 commits out of 2000: https://github.com/ammaarreshi/Generals-Mac-iOS-iPad/commits...
133.
by bel8 3mo ago
That's what VSCode next edit does too. It gathers context from the surroundings and recent actions to suggest blocks of code. It often feels magical. But I find agentic plan->review->implement->review workflow to be a net posi
134.
by bel8 3mo ago
I can speak for Chrome's MCP which I have more experience. devtools MCP will have access to more deep level fetures such as performance profiling, lighthouse and network requests in details (headers, auth, cookies...). For example, I h
135.
by bel8 3mo ago
Yeah it's probably some free or entry level LLM service. Even DeepSeek v4 Flash has 1million context size.
136.
by bel8 3mo ago
I don't think China would strugle to scrape the internet for fresh data. And they constantly publish state of the art LLM research (see DS4 context compaction and cache tech). They have very capable tech giants. So while not being able
137.
by bel8 3mo ago
You're on a good path. While consulting, I have seen .NET + htmx in production, but it was an internal tool used by thousands of folks (HR system). The 1ms response times and quick browser rendering due to no JS bloat put a smile on my
138.
by bel8 3mo ago
I expected Bun to be among the fastest and yes, I was not surprised.
139.
by bel8 3mo ago
Yes I've experienced everything you stated. Here's what helped me: Problem 1: "Obsessed with reinventing the wheel" " three duplicate functions": Suggestion: plan then implement. Tell LLM to scan your project a
140.
by bel8 3mo ago
Looks like a DSL for UI. Interesting. I entered "CRUD for public book library" in their playground ( https://www.openui.com/playground ) and it generated some forms which looks useful: https://i.imgur.com
141.
by bel8 3mo ago
I have been using Chrome's official MCP devtools server since Nov 2025. https://github.com/ChromeDevTools/chrome-devtools-mcp Before that I used Chrome web drivers but MCP is faster and more capable. I also instru
142.
by bel8 3mo ago
It's all great to leverage until something breaks in a middle layer and you can't reproduce without submitting your entire project in the GitHub issue.
143.
by bel8 3mo ago
They probably already have for internal use only. Unlike Anthropic and OpenAI, Google is much larger than AI. They don't need to sell SOTA.
144.
by bel8 3mo ago
For DS4 it's much cheaper and reputable to use OpenCode Go $10/mo subscription, or directly with DeepSeek API.
145.
by bel8 3mo ago
That's impressive! A quick glance shows me there's a commit that added almost 2500 lines of code. https://github.com/golemui/golemui/commit/861b182556fab44d1c... But that doesn't mean vibe-code
146.
by bel8 3mo ago
There's nothing sexy about it. From the get go they were accused of FUD drama and that it backfired as predicted. Now what I see in reality is that Anthropic mostly reaped more resentment from their poor marketing attempt. And their mo
147.
by bel8 3mo ago
> OpenAI would kill to get that. I have news for you, from 5 days ago. https://news.ycombinator.com/item?id=48690101 "U.S. government will decide who gets to use GPT-5.6"
148.
by bel8 3mo ago
If Opus 5 is capable enough, it will have the same guardrails with blocks/downgrades as Fable. So now we are supposed to cheer for Opus 5 to be just a mild improvement at best?
149.
by bel8 3mo ago
And what happens if you also get blocked by Opus? I couldn't find if they refund the tokens for that session or the customer is SOL.
150.
by bel8 3mo ago
That's something that Sonnet 5 couldn't get right even on a static SVG.

More ›