<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: bazlightyear</title><link>https://news.ycombinator.com/user?id=bazlightyear</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 05 May 2026 08:25:20 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=bazlightyear" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by bazlightyear in "Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge"]]></title><description><![CDATA[
<p>BTW it looks like Kimi won the subsequent challenge too <a href="https://aicc.rayonnant.ai/challenges/hexquerques/" rel="nofollow">https://aicc.rayonnant.ai/challenges/hexquerques/</a></p>
]]></description><pubDate>Sun, 03 May 2026 06:06:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=47993864</link><dc:creator>bazlightyear</dc:creator><comments>https://news.ycombinator.com/item?id=47993864</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47993864</guid></item><item><title><![CDATA[New comment by bazlightyear in "Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge"]]></title><description><![CDATA[
<p>Are you tests and results open source?</p>
]]></description><pubDate>Sun, 03 May 2026 05:39:44 +0000</pubDate><link>https://news.ycombinator.com/item?id=47993729</link><dc:creator>bazlightyear</dc:creator><comments>https://news.ycombinator.com/item?id=47993729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47993729</guid></item><item><title><![CDATA[Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge]]></title><description><![CDATA[
<p>Article URL: <a href="https://thinkpol.ca/2026/04/30/an-open-weights-chinese-model-just-beat-claude-gpt-5-5-and-gemini-in-a-programming-challenge/">https://thinkpol.ca/2026/04/30/an-open-weights-chinese-model-just-beat-claude-gpt-5-5-and-gemini-in-a-programming-challenge/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47993235">https://news.ycombinator.com/item?id=47993235</a></p>
<p>Points: 375</p>
<p># Comments: 218</p>
]]></description><pubDate>Sun, 03 May 2026 04:05:28 +0000</pubDate><link>https://thinkpol.ca/2026/04/30/an-open-weights-chinese-model-just-beat-claude-gpt-5-5-and-gemini-in-a-programming-challenge/</link><dc:creator>bazlightyear</dc:creator><comments>https://news.ycombinator.com/item?id=47993235</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47993235</guid></item><item><title><![CDATA[Claude vs. Gemini vs. ChatGPT vs. Grok]]></title><description><![CDATA[
<p>Article URL: <a href="https://boreal.social/post/i-pitted-claude-gemini-chatgpt-and-grok-against-each-other">https://boreal.social/post/i-pitted-claude-gemini-chatgpt-and-grok-against-each-other</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47449603">https://news.ycombinator.com/item?id=47449603</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 20 Mar 2026 02:15:57 +0000</pubDate><link>https://boreal.social/post/i-pitted-claude-gemini-chatgpt-and-grok-against-each-other</link><dc:creator>bazlightyear</dc:creator><comments>https://news.ycombinator.com/item?id=47449603</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47449603</guid></item><item><title><![CDATA[Show HN: GreedyPhrase – 1.21x better compression than GPT-4o tiktoken, 6x faster]]></title><description><![CDATA[
<p>A greedy phrase-based tokenizer that outperforms GPT-4 and GPT-4o tokenizers on compression,
with a smaller vocabulary.<p>Benchmark (enwik9, 1 GB):<p>Tokenizer                    Vocab Size  Total Tokens  Ratio   Throughput
GreedyPhrase                 65,536      222,805,405   4.49x   47 MB/s
Tiktoken o200k_base (GPT-4o) 200,019     270,616,861   3.70x   4.35 MB/s  
Tiktoken cl100k_base (GPT-4) 100,277     273,662,103   3.65x   7.13 MB/s<p>GreedyPhrase: 1.23x better than GPT-4, 1.21x better than GPT-4o. 1.5-3x smaller vocab,
6-11x higher encoding throughput.<p>How It Works:<p>1. Phrase Mining — Split into atoms (words, punctuation, whitespace). Mine bigrams/trigrams.
Top phrases fill 95% vocab slots.<p>2. BPE Fallback — Train BPE on residual byte sequences. Fills remaining 5% vocab.<p>3. Greedy Encoding — Longest-match-first Trie. Byte fallback for unknowns (zero OOV).</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47055617">https://news.ycombinator.com/item?id=47055617</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 18 Feb 2026 00:55:01 +0000</pubDate><link>https://github.com/rayonnant-ai/greedyphrase</link><dc:creator>bazlightyear</dc:creator><comments>https://news.ycombinator.com/item?id=47055617</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47055617</guid></item></channel></rss>