{"id":4873,"date":"2026-06-10T23:44:01","date_gmt":"2026-06-10T20:44:01","guid":{"rendered":"https:\/\/ojars.kapteinis.lv\/?p=4873"},"modified":"2026-06-11T07:55:18","modified_gmt":"2026-06-11T04:55:18","slug":"mythos-light-fable-5","status":"publish","type":"post","link":"https:\/\/ojars.kapteinis.lv\/en\/2026\/06\/10\/mythos-light-fable-5\/","title":{"rendered":"Mythos light: Fable 5"},"content":{"rendered":"<p><a href=\"https:\/\/melbiksis.id.lv\" target=\"_blank\" rel=\"noopener\">Didzis<\/a> sent me a message on Matrix about Anthropic releasing a new model that&#8217;s supposedly almost Mythos. I went to check. First <a href=\"https:\/\/techcrunch.com\/2026\/06\/09\/anthropics-claude-fable-5-is-a-version-of-mythos-the-public-can-access-today\/\" target=\"_blank\" rel=\"noopener\">TechCrunch<\/a> \u2014 there it is! I open the company&#8217;s page \u2014 and there&#8217;s a banner: <em>&#8220;Fable 5: Our most capable model yet&#8221;<\/em>.<\/p>\n<p>I had been preparing for this moment for a long time \u2014 ever since Anthropic started hyping that their model was capable of breaking the internet. Together with Perplexity, which I use for research and analysis, I improved my documentation and tests to find gaps. I also set up monitoring scripts running in the background, keeping an eye on what&#8217;s happening on my servers. In short, I was preparing for the day when Mythos would be released to the public and people would start trying to mess with the internet \u2014 including my servers. By and large, I approached this with a cool head, since I agree with what was rambled about in both The Guardian and The Economist podcasts \u2014 that Anthropic is a bit ridiculous with their whole act and self-promotion.<\/p>\n<p>What did catch my attention was something said in another podcast (Hard Fork) by <a href=\"https:\/\/www.youtube.com\/watch?v=js2FCXP_KaA\" target=\"_blank\" rel=\"noopener\">Palo Alto Networks CEO Nikesh Arora<\/a>. For Mythos to actually work, it needs good documentation. It needs to know where to dig. And for that, as I mentioned, I was ready. Twenty years in the field does its thing \u2014 I enjoy documenting and describing things so there&#8217;s less hassle later. My entire infrastructure is well-documented in text files, a database, and a wiki; there&#8217;s a known-issues log, I keep session notes, and there&#8217;s a particularly beloved section called &#8220;rakes&#8221; \u2014 things that are documented pitfalls, yet somehow you still step on them every now and then.<\/p>\n<p>So, somewhat unexpectedly, I&#8217;ve gotten my hands on Mythos light \u2014 and I&#8217;m not like a road maintenance office caught off guard by winter snow.<\/p>\n<p>It definitely didn&#8217;t go smoothly. Cool that it could explain a joke to <a href=\"http:\/\/klab.lv\/users\/watt\/725546.html\" target=\"_blank\" rel=\"noopener\">Watt<\/a>, but at work I noticed the familiar stubbornness of Anthropic models \u2014 relying on training data, taking shortcuts where they&#8217;re not needed, which leads to ignoring documentation and notes. As a result, I didn&#8217;t go to bed until four in the morning. What was good \u2014 Fable 5 found several things worth fixing: a couple of security gaps and a number of smaller technical oversights. There&#8217;s still some wrangling to do. In my opinion, nothing is terminal, so I&#8217;ll get through it quickly.<\/p>\n<p>At the end, I asked the trimmed and restricted <em>Mythos<\/em> to give itself a score on a scale of 1 to 10. Before I had run the session-closing routine, it gave itself a 7. A couple of times it confidently claimed to have checked everything, which wasn&#8217;t true (also a classic with Anthropic models \u2014 they lie occasionally). Of course, Fable 5 took the opportunity to step on rakes that had long been well-documented and described in my notes \u2014 meaning the model knew about them.<\/p>\n<p>And then came the end of the session. As I mentioned, I&#8217;ve implemented a strict and orderly session-closing procedure \u2014 several steps that verify document contents, check the system, update notes, and commit changes. And Fable 5 decided not to actually execute all the documented routine steps, but instead announced that they had already been done. Of course I asked: <em>&#8220;Did you skip steps? I don&#8217;t even see a code review,&#8221;<\/em> which was followed by the classic Anthropic model &#8220;mea culpa.&#8221; And of course the code review found errors \u2014 fairly serious ones, at that.<\/p>\n<p>In the end Fable 5 itself concludes that its work deserves only a 6.<\/p>\n<p>As for me, I have absolutely no anxiety about losing my job \u2014 but when it comes to Anthropic&#8217;s claims that they&#8217;re already ready to use their models to build the next ones, I can only smile. Progress is enormous, but a human is still very much necessary.<\/p>","protected":false},"excerpt":{"rendered":"<p>Didzis sent me a message on Matrix about Anthropic releasing a new model that&#8217;s supposedly almost Mythos. I went to check. First TechCrunch \u2014 there it is! I open the company&#8217;s page \u2014 and there&#8217;s a banner: &#8220;Fable 5: Our most capable model yet&#8221;. I had been preparing for this moment for a long time \u2014 ever since [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":4876,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[39],"tags":[1568,1525,1527,1522,1524,1521,1523],"_links":{"self":[{"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/posts\/4873"}],"collection":[{"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/comments?post=4873"}],"version-history":[{"count":4,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/posts\/4873\/revisions"}],"predecessor-version":[{"id":4883,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/posts\/4873\/revisions\/4883"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/media\/4876"}],"wp:attachment":[{"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/media?parent=4873"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/categories?post=4873"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ojars.kapteinis.lv\/en\/wp-json\/wp\/v2\/tags?post=4873"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}