<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: samvher</title><link>https://news.ycombinator.com/user?id=samvher</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 03:53:08 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=samvher" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by samvher in "Ask HN: What's your current AI agentic setup?"]]></title><description><![CDATA[
<p>I use it in a devcontainer. Protects me from most agent stupidity, though not malice. I give it read-only access to git origin and relevant AWS infra, so any remote changes are made by me from the host (manually or scripted), but it is still able to quickly get feedback on changes and look at logs.</p>
]]></description><pubDate>Thu, 05 Nov 2026 14:04:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=50379206</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=50379206</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50379206</guid></item><item><title><![CDATA[New comment by samvher in "DeepSeek v4.1 Flash Is Now Our Best Hacking Model"]]></title><description><![CDATA[
<p>I've been testing these models with fireworks.ai<p>They provide serverless access, and if you use fireconnect it's also easy to hook the models up to e.g. codex which allows for fairly clean comparisons<p>I think they provide the bare model (no filter) but I might be wrong, haven't tried using it for security research yet</p>
]]></description><pubDate>Wed, 16 Sep 2026 18:05:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=49730735</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=49730735</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49730735</guid></item><item><title><![CDATA[New comment by samvher in "GPT‑Live‑1 in the API"]]></title><description><![CDATA[
<p>Have you been able to find information on which languages are explicitly supported by this API? I'm having a hard time finding that out.</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:41:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656755</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=49656755</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656755</guid></item><item><title><![CDATA[New comment by samvher in "Resist "AI""]]></title><description><![CDATA[
<p>Yes and many forms of electricity generation evaporate water as well</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:28:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656618</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=49656618</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656618</guid></item><item><title><![CDATA[New comment by samvher in "Tell HN: OpenAI keeps re-enabling the 'allow training' setting"]]></title><description><![CDATA[
<p>I also noticed this on 2 accounts - did not take as careful notes as you did unfortunately. But I'm fairly confident - both of these are accounts where I care about the interactions not being used for training.<p>What's kind of still an open question for me is if the toggle automatically also applies to my Codex CLI use on the same account, or if that data is still silently being used in some way.<p>After this happened, I deleted all my ChatGPT history (even though I'm not sure how much it helps at this point), but for Codex I still haven't really found any way to do the same, I can still load my past sessions even after archiving them.</p>
]]></description><pubDate>Thu, 10 Sep 2026 15:14:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=49645177</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=49645177</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49645177</guid></item><item><title><![CDATA[Stealing Reasoning Traces from Proprietary LLM APIs]]></title><description><![CDATA[
<p>Article URL: <a href="https://arxiv.org/abs/2608.09867">https://arxiv.org/abs/2608.09867</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49259799">https://news.ycombinator.com/item?id=49259799</a></p>
<p>Points: 9</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 11 Aug 2026 15:23:14 +0000</pubDate><link>https://arxiv.org/abs/2608.09867</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=49259799</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49259799</guid></item><item><title><![CDATA[New comment by samvher in "How to burst the AI bubble: Strike at its roots"]]></title><description><![CDATA[
<p>Interesting challenge.<p>1 - I worked abroad and wasn't really familiar with the systems there. Gemini made me aware of a kind of pension account that I could withdraw from when I left the country netting me a few thousand dollars.<p>2 - Working as a tech contractor, charging by deliverable, Codex/Claude Code speed me up and it doesn't seem to have significantly dropped rates in the market.<p>3 - Also contractor related: I had Claude do a quick legal sanity check of my contracts, and it warned me of some clauses that I'd be better off removing/changing/refining. I was not aware of these nuances and would not have paid a lawyer for this as the contract was too small, but the changes were accepted by the client and reduced my risk exposure meaningfully.<p>4 - Learning a foreign language, I use it to check my draft emails and messages. It corrects them but also serves a tutoring role providing feedback, improving both the accuracy of my communication and my rate of language acquisition.<p>5 - Gemini Deep Research helped me narrow down tent models that met my fairly specific set of requirements. Very happy with the tent I ended up buying, from a brand that was not on my radar before.</p>
]]></description><pubDate>Wed, 24 Jun 2026 11:46:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=48658311</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=48658311</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48658311</guid></item><item><title><![CDATA[New comment by samvher in "What we know about the plan to give Americans an equity stake in AI"]]></title><description><![CDATA[
<p><a href="https://archive.ph/ainRf" rel="nofollow">https://archive.ph/ainRf</a></p>
]]></description><pubDate>Mon, 08 Jun 2026 04:09:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=48441237</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=48441237</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48441237</guid></item><item><title><![CDATA[What we know about the plan to give Americans an equity stake in AI]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.ft.com/content/8559a3f9-86de-4a1c-8a75-6623e83e6a00">https://www.ft.com/content/8559a3f9-86de-4a1c-8a75-6623e83e6a00</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=48441234">https://news.ycombinator.com/item?id=48441234</a></p>
<p>Points: 4</p>
<p># Comments: 2</p>
]]></description><pubDate>Mon, 08 Jun 2026 04:08:39 +0000</pubDate><link>https://www.ft.com/content/8559a3f9-86de-4a1c-8a75-6623e83e6a00</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=48441234</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48441234</guid></item><item><title><![CDATA[New comment by samvher in "Making 2.5 Flash and 2.5 Pro GA, and introducing Gemini 2.5 Flash-Lite"]]></title><description><![CDATA[
<p>Yes I was very surprised after the whole "scandal" around ChatGPT becoming too sycophantic that there was this massive change in tone from the last preview model (05-06) to the 06-05/GA model. The tone is really off-putting, I really liked how the preview versions felt like intelligent conversation partners and recognize what you're saying about useful pushback - it was my favorite set of models (the few preview iterations before this one) and I'm sad to see them disappearing.<p>Many people on the Google AI Developer forums have also noted either bugs or just performance regression in the final model.</p>
]]></description><pubDate>Wed, 18 Jun 2025 15:01:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=44310520</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=44310520</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44310520</guid></item><item><title><![CDATA[New comment by samvher in "AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms"]]></title><description><![CDATA[
<p><a href="https://ppc.cs.aalto.fi/" rel="nofollow">https://ppc.cs.aalto.fi/</a> covers some of this (overlapping with the topics the person you responded to mentioned, but not covering all, and including some others)</p>
]]></description><pubDate>Thu, 15 May 2025 15:11:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=43995871</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43995871</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43995871</guid></item><item><title><![CDATA[New comment by samvher in "Gemini Live with camera and screen sharing capabilities"]]></title><description><![CDATA[
<p>If the only moat really will be the scale of computation resources, that's great news for users, because it will be an extremely competitive market where prices will be driven down very effectively.<p>I suspect that model quality/vibes and integrations will play a role as well though.</p>
]]></description><pubDate>Fri, 11 Apr 2025 15:40:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=43655104</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43655104</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43655104</guid></item><item><title><![CDATA[New comment by samvher in "All clothing is handmade (2022)"]]></title><description><![CDATA[
<p>How did you go about learning this? It's something that I've been thinking I might be interested in picking up.</p>
]]></description><pubDate>Wed, 26 Mar 2025 15:42:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=43483433</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43483433</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43483433</guid></item><item><title><![CDATA[New comment by samvher in "DeepSeek Open Source FlashMLA – MLA Decoding Kernel for Hopper GPUs"]]></title><description><![CDATA[
<p>I can't tell if you're insinuating that Singapore is a pass-through for H100's heading towards China or whether there is some significant development taking place in Singapore that I'm unaware of?</p>
]]></description><pubDate>Mon, 24 Feb 2025 13:36:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=43159397</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43159397</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43159397</guid></item><item><title><![CDATA[New comment by samvher in "Grok3 Launch [video]"]]></title><description><![CDATA[
<p>What's happening definitely makes me nervous, but "at best a WW3-event and at worst an extinction-event" seems a bit much. Mainly because there are a _lot_ of unknowns. Better try to get comfortable with just riding this out.</p>
]]></description><pubDate>Tue, 18 Feb 2025 16:44:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=43091861</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43091861</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43091861</guid></item><item><title><![CDATA[New comment by samvher in "Matt Levine: Sure Elon Musk Might Buy OpenAI"]]></title><description><![CDATA[
<p><a href="https://archive.ph/fK0Tr" rel="nofollow">https://archive.ph/fK0Tr</a></p>
]]></description><pubDate>Tue, 11 Feb 2025 17:47:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=43015750</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43015750</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43015750</guid></item><item><title><![CDATA[Matt Levine: Sure Elon Musk Might Buy OpenAI]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.bloomberg.com/opinion/articles/2025-02-11/sure-elon-musk-might-buy-openai">https://www.bloomberg.com/opinion/articles/2025-02-11/sure-elon-musk-might-buy-openai</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=43015744">https://news.ycombinator.com/item?id=43015744</a></p>
<p>Points: 8</p>
<p># Comments: 1</p>
]]></description><pubDate>Tue, 11 Feb 2025 17:47:27 +0000</pubDate><link>https://www.bloomberg.com/opinion/articles/2025-02-11/sure-elon-musk-might-buy-openai</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=43015744</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43015744</guid></item><item><title><![CDATA[New comment by samvher in "The New York Times Has Spent $10.8M in Its Legal Battle with OpenAI So Far"]]></title><description><![CDATA[
<p>I'm trying to parse the idea of "a collective agreement" but can't fully wrap my head around how that would work.<p>It seems to me more like the lack of a "Walmart Law" is a result of e.g. lack of economies of scale and other economic structure, rather than some collective agreement. (If it was profitable to break out of that agreement and start a "Walmart Law", it seems we'd see that happen pretty quickly?)<p>But if you know more about this and I'm off the mark I'd love to learn</p>
]]></description><pubDate>Wed, 05 Feb 2025 19:13:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=42953578</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=42953578</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42953578</guid></item><item><title><![CDATA[New comment by samvher in "DeepSeek R1 Is Now Available on Azure AI Foundry and GitHub"]]></title><description><![CDATA[
<p>For me the excitement is that around the o3 announcement I had a feeling like we were heading to an OpenAI / Sam Altman controlled dystopia. This resets that - you can run the model yourself, you can modify it yourself, it's essentially on par with the best public models, and it gives hope that the smaller players have a fighting chance going forward. They also published their innovations bringing back some of the feeling of open science that used to be in ML research but which mostly went away.</p>
]]></description><pubDate>Wed, 29 Jan 2025 21:53:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=42871701</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=42871701</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42871701</guid></item><item><title><![CDATA[New comment by samvher in "Complete hardware and software setup for running Deepseek-R1 locally"]]></title><description><![CDATA[
<p>Given what we just saw in terms of the DeepSeek team squeezing a lot of extra performance out of more efficient implementation on GPU, and the model still being optimized for GPU rather than CPU - is it unreasonable to think that in the $6k setup described, some performance might still be left on the table that could be squeezed out with some better optimization for these particular CPUs?</p>
]]></description><pubDate>Wed, 29 Jan 2025 16:57:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=42867650</link><dc:creator>samvher</dc:creator><comments>https://news.ycombinator.com/item?id=42867650</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42867650</guid></item></channel></rss>