<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: msp26</title><link>https://news.ycombinator.com/user?id=msp26</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 19 Aug 2026 18:57:35 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=msp26" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by msp26 in "Qwen3.8 Max now ranked as the best overall model by agentic index"]]></title><description><![CDATA[
<p>Not sure how to fully fix this but I remember a session last week where I got so fed up mid way though reading a response that I used the following:<p>"give me this again without jargon invented this session at high density<p>and with a couple (maybe more or less) simple useful ascii diagrams underneath each design"<p>The context is that I was discussing an experimental new idea for my video game review analysis product.<p>Designs 1,2, and 3 were horrible: the model even suggested a rejection after the word soup so it would have been pointless to waste my fleeting time on Earth reading it.<p>Otherwise, I generally really enjoyed using fable for bouncing ideas. It was an absolute joy to have this thing provide useful criticism, analyse sample data, and create prototypes so that I could elevate my understanding of the problem without stepping down from a pure intuition/design headspace.<p>But I don't consider the purely model written code usable for a feature this important. I'll probably scrap it entirely and start from scratch with newfound understanding.</p>
]]></description><pubDate>Thu, 06 Aug 2026 21:53:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=49203129</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49203129</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49203129</guid></item><item><title><![CDATA[New comment by msp26 in "Qwen3.8 Max now ranked as the best overall model by agentic index"]]></title><description><![CDATA[
<p>No the models are just ass at communication without being directed.<p>Try asking them to make useful diagrams for some stuff in a codebase, out of the box without excessive hand holding they don't make good choices about what's worth communicating and how to do it.<p>You see this in their pointless frontend copy all the time too.</p>
]]></description><pubDate>Thu, 06 Aug 2026 21:20:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49202738</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49202738</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49202738</guid></item><item><title><![CDATA[New comment by msp26 in "Qwen3.8 Max now ranked as the best overall model by agentic index"]]></title><description><![CDATA[
<p>yep matches my experience completely<p>But even fable has the annoying tendency to invent new jargon and produce an incomprehensible soup of text.</p>
]]></description><pubDate>Thu, 06 Aug 2026 21:17:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=49202695</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49202695</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49202695</guid></item><item><title><![CDATA[New comment by msp26 in "Predictive Speculative KV Replication for Bursty LLM Inference"]]></title><description><![CDATA[
<p>the hand drawn diagrams and highlights are charming</p>
]]></description><pubDate>Fri, 31 Jul 2026 22:58:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=49129460</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49129460</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49129460</guid></item><item><title><![CDATA[New comment by msp26 in "Opus 5 is currently #1 on Artificial Analysis Intelligence Leaderboard"]]></title><description><![CDATA[
<p>Asking fable to read it's own model card triggers this btw. Or asking if mitochondria is the powerhouse of the cell.</p>
]]></description><pubDate>Fri, 24 Jul 2026 23:45:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=49042984</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49042984</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49042984</guid></item><item><title><![CDATA[New comment by msp26 in "Marimo now runs in PyCharm"]]></title><description><![CDATA[
<p>Last I heard, the feature was in beta so avoided it. But I'll definitely give it a go if it's mature now!<p>I have been using the --watch flag to let my agent play with the notebook as I use it already.</p>
]]></description><pubDate>Fri, 24 Jul 2026 23:26:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=49042850</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49042850</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49042850</guid></item><item><title><![CDATA[New comment by msp26 in "Marimo now runs in PyCharm"]]></title><description><![CDATA[
<p>I do that a decent chunk of the time yeah especially for learning. I also have a bunch of marimo notebooks that double as clis and they're lovely.<p>But sometimes I want to do something too specific or high fidelity and it's just easier to get the clanker to write typescript and make a webpage/components.</p>
]]></description><pubDate>Fri, 24 Jul 2026 23:23:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=49042834</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49042834</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49042834</guid></item><item><title><![CDATA[New comment by msp26 in "Marimo now runs in PyCharm"]]></title><description><![CDATA[
<p>I fucking love marimo for exploring data.<p>However my use of it has decreased a little with how easily I can conjure disposable frontends with agents to explore one off things.</p>
]]></description><pubDate>Fri, 24 Jul 2026 21:56:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=49042093</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=49042093</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49042093</guid></item><item><title><![CDATA[New comment by msp26 in "Claude Code: Anatomy of a Misfeature"]]></title><description><![CDATA[
<p>hello, please fix needing to reauth every day (sometimes with email verification). This started happening this month. It's tedious and makes switching very tempting.<p>I'm using the VS Code extension over SSH.</p>
]]></description><pubDate>Fri, 17 Jul 2026 17:17:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=48949790</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48949790</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48949790</guid></item><item><title><![CDATA[New comment by msp26 in "Fable extended until 19 July"]]></title><description><![CDATA[
<p>Deeply unserious company, flip flopping on policy every week with ludicrous, sometimes invisible, guard rails on their top model. Imagine trying to make business decisions about AI use with this.<p>I love Fable for many things including coding but I cannot justify basing any internal AI tools or LLM powered products on top of Anthropic's offerings.</p>
]]></description><pubDate>Sun, 12 Jul 2026 18:18:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=48883206</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48883206</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48883206</guid></item><item><title><![CDATA[New comment by msp26 in "The text in Claude Code’s “Extended Thinking” output"]]></title><description><![CDATA[
<p>> Summarized thinking provides the full intelligence benefits of extended thinking, while preventing misuse.<p>> preventing misuse.<p>Imagine not being able to read the tokens you are paying for.</p>
]]></description><pubDate>Mon, 22 Jun 2026 16:18:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=48632253</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48632253</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48632253</guid></item><item><title><![CDATA[New comment by msp26 in "Running local models is good now"]]></title><description><![CDATA[
<p>Yep agreed completely. I couldn't imagine torturing myself with a small model for local coding. But Gemma 4 31B is so fucking good for a variety of language modelling tasks.</p>
]]></description><pubDate>Tue, 16 Jun 2026 17:51:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=48559146</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48559146</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48559146</guid></item><item><title><![CDATA[New comment by msp26 in "Claude Fable 5"]]></title><description><![CDATA[
<p>It triggered for me when I asked "Web search for your own model card (released today) and pick out your favourite highlights from the pdf"</p>
]]></description><pubDate>Tue, 09 Jun 2026 18:40:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=48465592</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48465592</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48465592</guid></item><item><title><![CDATA[New comment by msp26 in "Claude Fable 5"]]></title><description><![CDATA[
<p>>Pricing for both models is $10 per million input tokens and $50 per million output tokens.</p>
]]></description><pubDate>Tue, 09 Jun 2026 17:08:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=48463978</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48463978</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48463978</guid></item><item><title><![CDATA[New comment by msp26 in "I’ve joined Anthropic"]]></title><description><![CDATA[
<p>hell will freeze over before anthropic release anything meaningful to the public</p>
]]></description><pubDate>Tue, 19 May 2026 16:44:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=48195767</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48195767</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48195767</guid></item><item><title><![CDATA[New comment by msp26 in "Accelerating Gemma 4: faster inference with multi-token prediction drafters"]]></title><description><![CDATA[
<p>Interesting, I might try that, thanks!</p>
]]></description><pubDate>Wed, 06 May 2026 08:13:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=48033628</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48033628</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48033628</guid></item><item><title><![CDATA[New comment by msp26 in "Accelerating Gemma 4: faster inference with multi-token prediction drafters"]]></title><description><![CDATA[
<p>Google is singlehandedly carrying western open source models. Gemma 4 31B is fantastic.<p>However, it is a little painful to try to fit the best possible version into 24GB vram with vision + this drafter soon. My build doesn't support any more GPUs and I believe I would want another 4090 (overpriced) for best performance or otherwise just replace it altogether.</p>
]]></description><pubDate>Tue, 05 May 2026 18:06:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=48026231</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=48026231</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48026231</guid></item><item><title><![CDATA[New comment by msp26 in "Show HN: Mljar Studio – local AI data analyst that saves analysis as notebooks"]]></title><description><![CDATA[
<p>I like starting most of my projects on marimo notebooks now and slowly moving parts of it to the main codebase + db.<p>By the end of it I might remove the notebook entirely but usually I keep it for some visualisation + running stuff as a cli tool.</p>
]]></description><pubDate>Sat, 02 May 2026 16:12:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=47987662</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=47987662</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47987662</guid></item><item><title><![CDATA[New comment by msp26 in "Claude.ai unavailable and elevated errors on the API"]]></title><description><![CDATA[
<p>session usage limits this week feel like ass. Even when being careful to not break prefix caching.</p>
]]></description><pubDate>Tue, 28 Apr 2026 18:20:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=47938340</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=47938340</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47938340</guid></item><item><title><![CDATA[New comment by msp26 in "Claude Token Counter, now with model comparisons"]]></title><description><![CDATA[
<p>Not necessarily with speculative decoding. Whitespace would be trivial to predict and they would petty much keep using the same amount of compute as before.<p>I don't think that's their primary motive for doing this but it is a side effect.</p>
]]></description><pubDate>Mon, 20 Apr 2026 10:11:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=47832277</link><dc:creator>msp26</dc:creator><comments>https://news.ycombinator.com/item?id=47832277</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47832277</guid></item></channel></rss>