<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: TobTobXX</title><link>https://news.ycombinator.com/user?id=TobTobXX</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Thu, 29 Oct 2026 13:21:55 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=TobTobXX" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by TobTobXX in "Gemini 4 Carbon"]]></title><description><![CDATA[
<p>> Gemini Carbon<p>> [Header image is a chunk of ice]<p>???</p>
]]></description><pubDate>Wed, 28 Oct 2026 18:47:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=50278940</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=50278940</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50278940</guid></item><item><title><![CDATA[New comment by TobTobXX in "Introducing LFM3: Every Modality for Specialized Intelligence"]]></title><description><![CDATA[
<p>Is LFM3-Omni-2B worse or roughly equal at vision tasks compared to LFM3-VL-2B?<p>In other words: Should I select the model that matches my desired capability set or just use Omni?</p>
]]></description><pubDate>Tue, 27 Oct 2026 23:24:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=50267304</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=50267304</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50267304</guid></item><item><title><![CDATA[New comment by TobTobXX in "Kimi K3.1 Pro and K3.1 Pro Highspeed"]]></title><description><![CDATA[
<p>I think that this is Kimi K3.1 Pro on max too. Unfortunately, AA doesn't seem to test for other effort levels for models not from Anthropic/OpenAI.</p>
]]></description><pubDate>Fri, 23 Oct 2026 20:13:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=50214665</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=50214665</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50214665</guid></item><item><title><![CDATA[New comment by TobTobXX in "Pi.dev: You Said No MCP"]]></title><description><![CDATA[
<p>> taking tool calls, code modifications, etc, and dumping them in favor of a summary<p>Doesn't that destroy the cache? I find that caching significantly sped up my Qwen, especially on said larger contexts.</p>
]]></description><pubDate>Wed, 30 Sep 2026 12:20:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=49907837</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=49907837</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49907837</guid></item><item><title><![CDATA[New comment by TobTobXX in "Ask HN: How do you manage skills files?"]]></title><description><![CDATA[
<p>A skill should only document behaviour the LLM didn't/couldn't exhibit on its own.<p>So you take your failed case (eg. working with gdb or whatever), write a skill and then test for that failed case.</p>
]]></description><pubDate>Mon, 07 Sep 2026 06:14:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=49594537</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=49594537</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49594537</guid></item><item><title><![CDATA[New comment by TobTobXX in "Run Qwen3.8 27B locally: real numbers from my Mac Studio"]]></title><description><![CDATA[
<p>Can you tell more about your experience with Ornith? I've come across it, and the benchmarks on its landing page are unbeleavably good. But it isn't featured on more established benchmarks like ArtificialAnalysis and it's not on OpenRouter, so I wrote it off as scam.</p>
]]></description><pubDate>Fri, 28 Aug 2026 17:50:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=49482086</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=49482086</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49482086</guid></item><item><title><![CDATA[New comment by TobTobXX in "Show HN: Reverse Minesweeper"]]></title><description><![CDATA[
<p>Isn't that just like Binoxxo? I've solved Binoxxo's every time I came across a newspaper.<p>Online example for the curious: <a href="https://raetselwelt.online/en/puzzles/binoxxo/" rel="nofollow">https://raetselwelt.online/en/puzzles/binoxxo/</a></p>
]]></description><pubDate>Sun, 26 Jul 2026 20:33:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=49062135</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=49062135</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49062135</guid></item><item><title><![CDATA[New comment by TobTobXX in "Codeberg Bans Cryptocurrency Projects"]]></title><description><![CDATA[
<p>No, I think both, political and apolitical decisions can exist.<p>If you ban a project because its CI pipeline mines crypto and that ia a drain on your server, it is an apolitical argument.<p>If you ban a project because it drains recources because an AI agent is commutting multiple times a second, that is apolitical.<p>If you ban a project, because it hosts content that doesn't align with your ideals since it uses cryptocurrencies, that is clearly political.<p>Sure, there may be a small grey zone. But there are swaths of decicions that are clearly political/apolitical.</p>
]]></description><pubDate>Thu, 23 Jul 2026 06:57:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49017830</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=49017830</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49017830</guid></item><item><title><![CDATA[New comment by TobTobXX in "EEG shows brain can simultaneous encode two speech streams"]]></title><description><![CDATA[
<p>That tracks. As a teacher I sometimes find myself conversing with multiple kids simultaniously as well. If it's nothing too deep that requires full focus, it works. (Though I do find it tiring and avoid it.)</p>
]]></description><pubDate>Fri, 17 Jul 2026 08:13:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=48944591</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48944591</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48944591</guid></item><item><title><![CDATA[New comment by TobTobXX in "Ancient Roman Board Game"]]></title><description><![CDATA[
<p>Too assymetric IMO. I have no problems winning with hounds even on hard, but I have no chance of winning except against easy.</p>
]]></description><pubDate>Mon, 13 Jul 2026 18:36:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=48896859</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48896859</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48896859</guid></item><item><title><![CDATA[New comment by TobTobXX in "Microsoft Can Track Users via a Windows Device ID"]]></title><description><![CDATA[
<p>Only side effect I ran into is that journalctl couldn't find the logs of a previous boot.</p>
]]></description><pubDate>Tue, 07 Jul 2026 20:42:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=48823451</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48823451</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48823451</guid></item><item><title><![CDATA[New comment by TobTobXX in "To have a moral stance on AI is to be an outcast, and it sucks"]]></title><description><![CDATA[
<p>I don't work in tech (school teacher), so the main way I interact with tech people is online.<p>IME, everyone I meet offline has some low-level caution about AI taking their job, but uses AI and is amazed by their capabilities and is glad for this tool.<p>Most ppl I meet online are strong anti-AI advocates.</p>
]]></description><pubDate>Sat, 30 May 2026 16:52:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=48338295</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48338295</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48338295</guid></item><item><title><![CDATA[New comment by TobTobXX in "Anthropic surpasses OpenAI to become most valuable AI startup"]]></title><description><![CDATA[
<p>Painting something new with great potential as negative and "the enemy" - this is truly the essence of humanity. </snark-but-with-a-kernel-of-truth><p>If the parent commenter enjoys working with LLMs, then just let them. This doesn't make them inhumane, nor does it make you less of a human.</p>
]]></description><pubDate>Sat, 30 May 2026 16:45:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=48338204</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48338204</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48338204</guid></item><item><title><![CDATA[New comment by TobTobXX in "Liquid AI reveals 8B-A1B MoE trained on 38T"]]></title><description><![CDATA[
<p>Ollama is just llama.cpp but with their own interface ontop. Liquid does support llama.cpp, but Ollama is slow in updating its llama.cpp dependency.</p>
]]></description><pubDate>Fri, 29 May 2026 19:31:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=48328108</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48328108</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48328108</guid></item><item><title><![CDATA[New comment by TobTobXX in "Show HN: Needle: We Distilled Gemini Tool Calling into a 26M Model"]]></title><description><![CDATA[
<p>Wait what? I've used DeepSeek V4 flash a lot and compsred to Gemma 4 E2B (ie. the smallest, event at q4), it consistently underperforms. In contrast to DS flash, I've found Gemma 4 to be incredibly precise and consistent with tool use.</p>
]]></description><pubDate>Wed, 13 May 2026 12:25:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=48121020</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48121020</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48121020</guid></item><item><title><![CDATA[New comment by TobTobXX in "Show HN: Needle: We Distilled Gemini Tool Calling into a 26M Model"]]></title><description><![CDATA[
<p><a href="https://asciinema.org/" rel="nofollow">https://asciinema.org/</a> in case anyome is searching it.</p>
]]></description><pubDate>Wed, 13 May 2026 12:20:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=48120965</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=48120965</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48120965</guid></item><item><title><![CDATA[New comment by TobTobXX in "Muse Spark: Scaling towards personal superintelligence"]]></title><description><![CDATA[
<p>> Muse Spark is a natively multimodal reasoning model with support for [...] visual chain of thought [...].<p>Do they mean "the chain of thought is visible to the user" (ie. not hidden like ChatGPT), or "the medium of the chain of thought is not text, but visuals" (ie. thinking in images).<p>I'd guess the former, since it wouldn't be economical to generate transient images, just for thinking. But I'm not sure why they'd highight that in that case. If it were the second thing, that'd be extremely interesting. The first model not to think in text.</p>
]]></description><pubDate>Wed, 08 Apr 2026 21:15:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=47696376</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=47696376</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47696376</guid></item><item><title><![CDATA[New comment by TobTobXX in "German implementation of eIDAS will require an Apple/Google account to function"]]></title><description><![CDATA[
<p>Somewhat. To fill out my taxes online, I could sign up with either the AGOV app (needs Google Android) or a USB security key. I happened to have a yubikey, but I needed to mess with the firefox about:config (security.webauth.u2f=true IIRC). It did work in the end though.</p>
]]></description><pubDate>Sun, 05 Apr 2026 17:31:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=47651748</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=47651748</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47651748</guid></item><item><title><![CDATA[New comment by TobTobXX in "German implementation of eIDAS will require an Apple/Google account to function"]]></title><description><![CDATA[
<p>Same in Switzerland. The app needed to sign in to fill out my taxes doesn't work on ungoogled Android.</p>
]]></description><pubDate>Sun, 05 Apr 2026 06:47:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=47646803</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=47646803</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47646803</guid></item><item><title><![CDATA[New comment by TobTobXX in "Artemis II Launch Day Updates"]]></title><description><![CDATA[
<p>330 cubic feet =~ 9.3m²</p>
]]></description><pubDate>Thu, 02 Apr 2026 08:49:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=47611725</link><dc:creator>TobTobXX</dc:creator><comments>https://news.ycombinator.com/item?id=47611725</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47611725</guid></item></channel></rss>