{"id":2247892,"date":"2026-11-09T19:24:51","date_gmt":"2026-11-09T16:24:51","guid":{"rendered":"https:\/\/analyse.optim.biz\/?p=2247892"},"modified":"2026-11-09T19:24:51","modified_gmt":"2026-11-09T16:24:51","slug":"anthropic-releases-new-mythos-and-fable-models-amid-rsi-concerns","status":"publish","type":"post","link":"https:\/\/analyse.optim.biz\/?p=2247892","title":{"rendered":"Anthropic releases new Mythos and Fable models amid RSI concerns"},"content":{"rendered":"<p>[analyse_image type=&#8221;featured&#8221; src=&#8221;https:\/\/techcrunch.com\/wp-content\/uploads\/2026\/06\/GettyImages-2279025960.jpg?w=1024&#8243;]<\/p>\n<div class=\"entry-content wp-block-post-content is-layout-constrained wp-block-post-content-is-layout-constrained\">\n<p id=\"speakable-summary\" class=\"wp-block-paragraph\">Anthropic released new versions of its most advanced models on Monday, with the launch of Mythos 5.2 and Fable 5.2. The company says the new models \u201cshow improvements in areas like coding, reasoning, and vision\u201d and communicate more clearly with users.<\/p>\n<p class=\"wp-block-paragraph\">Notably, Anthropic is lowering the cost of Fable from $10 to $8 per million input tokens. As a result, Fable 5.2 is now cheaper than its primary competitor GPT-6 Astra. Anthropic says that the new version of Fable will return responses 30% faster across effort levels. Faster inference serving is available for double the price.  <\/p>\n<p class=\"wp-block-paragraph\">Fable 5.2 is also compatible with zero data retention (ZDR) policies through the Enterprise Frontier Safeguards system, long a sticking point for many enterprise clients.<\/p>\n<p class=\"wp-block-paragraph\">It\u2019s Anthropic\u2019s first model release since the arrival of a third-party report from the assessment group METR, which suggested the lab might be approaching recursive self-improvement (RSI) \u2014 or the ability for AI models to carry on model training independently. The METR report found internal models were speeding up Anthropic\u2019s AI progress by roughly 2x, although many uncertainties remain.<\/p>\n<p class=\"wp-block-paragraph\">Still, Anthropic\u2019s release notes significant ongoing challenges in accurately testing model capabilities. \u201cClaude Mythos 5.2 and Fable 5.2 each voice suspicion that they are being evaluated more often than any of our recent models, which challenges our ability to assess how they will act in real-world settings. As these settings expand and capabilities increase with future models, we expect this challenge to grow.\u201d<\/p>\n<p class=\"wp-block-paragraph\">Anthropic maintains that the new version of Mythos does not have the ability to act covertly against its user, a verdict that\u2019s detailed further in Mythos 5.2\u2019s system card.<\/p>\n<p class=\"wp-block-paragraph\">\u201cClaude Mythos 5.2 shows stronger covert capabilities than prior Claude models on several of our evaluations, particularly when its reasoning is hidden from the monitor,\u201d the system card reads. \u201cThat said, we believe that low levels of absolute performance on our new covert evaluations support claims that rely on detection of harmful actions in high-stakes situations.\u201d<\/p>\n<p class=\"wp-block-paragraph\">Fable 5.2 has been available to select parties in pre-release for some time, and Anthropic says the model has already \u201csolved two open problems in theoretical physics, formalized Fermat\u2019s Last Theorem, and supported numerous other research projects.\u201d<\/p>\n<p class=\"wp-block-paragraph\">Also on Monday, AI startup Physical Superintelligence announced a new proof of the Aaronson\u2011Ambainis conjecture, a pivotal step in setting theoretical limits for quantum computing. Notably, the lab said the result was achieved by running \u201ca non-public Anthropic model\u201d within PSI\u2019s own system. Anthropic did not confirm or deny whether Fable 5.2 was the model in question.<\/p>\n<\/div>\n<p>[analyse_source url=&#8221;https:\/\/techcrunch.com\/2026\/11\/09\/anthropic-releases-new-mythos-and-fable-models-amid-rsi-concerns\/&#8221;]<\/p>\n","protected":false},"excerpt":{"rendered":"<p>[analyse_image type=&#8221;featured&#8221; src=&#8221;https:\/\/techcrunch.com\/wp-content\/uploads\/2026\/06\/GettyImages-2279025960.jpg?w=1024&#8243;] Anthropic released new versions of its most advanced models on Monday, with the launch of Mythos 5.2 and Fable 5.2. The company says the new models \u201cshow improvements in areas like coding, reasoning, and vision\u201d and communicate more clearly with users. Notably, Anthropic is lowering the cost of Fable from $10 to [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[226,62],"class_list":["post-2247892","post","type-post","status-publish","format-standard","hentry","category-politics","tag-crawlmanager","tag-techcrunch-com"],"_links":{"self":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2247892","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2247892"}],"version-history":[{"count":0,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=\/wp\/v2\/posts\/2247892\/revisions"}],"wp:attachment":[{"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2247892"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2247892"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/analyse.optim.biz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2247892"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}