{"id":7251,"date":"2026-09-07T12:27:40","date_gmt":"2026-09-07T12:27:40","guid":{"rendered":"https:\/\/justfineinfotech.com\/introducing-claude-fable-5-1-and-claude-mythos-5-1\/"},"modified":"2026-09-07T12:27:40","modified_gmt":"2026-09-07T12:27:40","slug":"introducing-claude-fable-5-1-and-claude-mythos-5-1","status":"publish","type":"post","link":"https:\/\/justfineinfotech.com\/fr\/introducing-claude-fable-5-1-and-claude-mythos-5-1\/","title":{"rendered":"Introducing Claude Fable 5.1 and Claude Mythos 5.1"},"content":{"rendered":"<figcaption>Terminal-Bench-Science 0.1Accuracy vs Cost<\/figcaption><ul>\n<li><strong>Fable 5.1<\/strong><\/li>\n<li><strong>Fable 5<\/strong><\/li>\n<\/ul>\n<p>Terminal-Bench-Science 0.1: The standard error is \u00b13.5\u20134.5 pts per model. The public leaderboard (3 trials\/task, <a href=\"https:\/\/justfineinfotech.com\/claude-ai-price-today-claude-live-price-chart-market-cap\/\" title=\"Claude Ai Price Today | Claude Live Price, Chart &amp; Market Cap\">Claude<\/a> Code harness) reports Claude Opus 5 at 30.0% and Claude Fable 5 at 21.4%; our setup reproduces them at 29.0% and 24.7%, respectively, both within noise<\/p><figcaption>Terminal-Bench 4.0Accuracy vs Cost<\/figcaption><ul>\n<li><strong>Mythos 5.1<\/strong><\/li>\n<li><strong>Fable 5.1<\/strong><\/li>\n<li><strong>Mythos 5<\/strong><\/li>\n<\/ul>\n<p>Terminal-Bench 4.0 scores by cost (log scale), at each effort level. Claude Fable 5.1 and Claude Mythos 5.1 are the same underlying model; the gap between them reflects the tasks on which our earlier, less precise cyber safeguards intervened. With the improvements we\u2019re making to these safeguards today, we expect the difference between the models to be much smaller<\/p><figcaption>Humanity&#8217;s Last ExamAccuracy vs Cost<\/figcaption><ul>\n<li><strong>Fable 5.1<\/strong> (with tools)<\/li>\n<li><strong>Fable 5.1<\/strong> (no tools)<\/li>\n<li><strong>Fable 5<\/strong> (with tools)<\/li>\n<li><strong>Fable 5<\/strong> (no tools)<\/li>\n<\/ul>\n<p>Humanity\u2019s Last Exam scores by cost (log scale), at each effort level. CursorBench 3.2.0 scores by cost (log scale), at each effort level<\/p><figcaption>CursorBench 3.2.0Accuracy vs Cost<\/figcaption><ul>\n<li><strong>Fable 5.1<\/strong><\/li>\n<li><strong>Fable 5<\/strong><\/li>\n<\/ul>\n<p>CursorBench 3.2.0 by cost (log scale), at each effort level<\/p>\n<table>\n<thead>\n<tr>\n<th>Fable 5.1<\/th>\n<th>Fable 5<\/th>\n<th>Opus 5<\/th>\n<th>GPT-5.6 Sol<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<th>Agentic scientific researchTerminal-Bench-Science 0.1 [1]<\/th>\n<td>52.6%<\/td>\n<td>24.7%<\/td>\n<td>29.0%<\/td>\n<td>22.4%<\/td>\n<\/tr>\n<tr>\n<th>Agentic codingTerminal-Bench 4.0<\/th>\n<td>55.8%60.9% (Mythos 5.1)<\/td>\n<td>42.0%<\/td>\n<td>52.3%<\/td>\n<td>37.3%<\/td>\n<\/tr>\n<tr>\n<th>Knowledge workGDPval-AA v2<\/th>\n<td>1853<\/td>\n<td>1723<\/td>\n<td>1824<\/td>\n<td>1711<\/td>\n<\/tr>\n<tr>\n<th>Computer useOSWorld 2.0 [2]<\/th>\n<td>77.9%partial<\/td>\n<td>72.9%partial<\/td>\n<td>75.4%partial<\/td>\n<td>\u2014partial<\/td>\n<\/tr>\n<tr>\n<th>Computer useOSWorld 2.0<\/th>\n<td>41.7%strict<\/td>\n<td>36.1%strict<\/td>\n<td>39.6%strict<\/td>\n<td>\u2014strict<\/td>\n<\/tr>\n<tr>\n<th>Multidisciplinary reasoningHumanity&#8217;s Last Exam<\/th>\n<td>60.9%no tools<\/td>\n<td>57.8%no tools<\/td>\n<td>56.6%no tools<\/td>\n<td>\u2014no tools<\/td>\n<\/tr>\n<tr>\n<td>65.0%with tools<\/td>\n<td>63.8%with tools<\/td>\n<td>63.6%with tools<\/td>\n<td>\u2014with tools<\/td>\n<\/tr>\n<tr>\n<th>Business workflowsAutomationBench<\/th>\n<td>31.4%<\/td>\n<td>17.1%<\/td>\n<td>26.9%<\/td>\n<td>19.6%<\/td>\n<\/tr>\n<tr>\n<th>Agentic codingCursorBench 3.2.0<\/th>\n<td>73.4%<\/td>\n<td>70.5%<\/td>\n<td>70.0%<\/td>\n<td>67.2%<\/td>\n<\/tr>\n<\/tbody>\n<\/table><figcaption>Fable 5.1 was evaluated with its production safeguards enabled. On tasks where these safeguards intervened, Fable 5.1 and Fable 5 scored a zero on OSWorld 2.0, and Fable 5 scored a zero on AutomationBench. In all other interventions from our safeguards, cybersecurity tasks were <a href=\"https:\/\/justfineinfotech.com\/social-media-management-with-claude-the-complete-guide\/\" title=\"Social Media Management with Claude: The Complete Guide\">complete<\/a>d by Claude Opus 4.8, and biology tasks were completed by Claude Opus 5. This likely reduces the performance of Fable 5.1 and Fable 5 on these benchmarks.<\/figcaption>Quote<\/p>\n<blockquote>\n<p>\u201cIn internal benchmarks, Claude Fable 5.1 solves more of our coding problems than Fable 5 or Opus 5, and achieves state of the art on trading intuition. While prior models became hard to follow the longer they worked, Fable 5.1 remains readable over long, multi-step tasks.\u201d<\/p>\n<\/blockquote>\n<p>CompanyJane Street Capital<br \/>\nAuthorCraig Falls, Head of Quantitative Research<br \/>\n01 \/ 22<figcaption>Claude-designed protein binders (orange) for each of 12 targets (grey). Every design in the video was confirmed to bind in the lab. Structures shown are ESMFold2 predictions.<\/figcaption><figure>\n<img decoding=\"async\" src=\"https:\/\/justfineinfotech.com\/wp-content\/uploads\/2026\/09\/image.webp\" alt=\"Radar image (Magellan): bright cone, radian lava flows\"><figcaption>Magellan radar<\/figcaption><\/figure><figcaption>Altimetry 10-20km footprint<\/figcaption><figcaption>New DEM (300m) a volcano 15km across<\/figcaption><figcaption>A small shield volcano on Venus<\/figcaption><figcaption>Inference speedup<\/figcaption><p>Inference speedup for seven open-<\/p><figcaption>Estimated cost savings on genome-wide analyses<\/figcaption><ul>\n<li>Original implementation<\/li>\n<li>Optimized<\/li>\n<\/ul>\n<p>Estimated GPU cost of three genome-wide analyses before and after optimization, at cloud list price. Evo 2 40B saves more on a whole job (2.3x) than per forward (1.4x) because some of its optimizations only pay off across many sequences<\/p><figcaption>Indexed cost of Fable usage<\/figcaption><ul>\n<li>Cache reads<\/li>\n<li>All other tokens<\/li>\n<\/ul>\n<p>Indexed cost of running the same workloads on Fable 5 and Fable 5.1, at usage-based pricing measured at default effort over four weeks of actual usage in August 2026. Typical workload covers Fable usage across Claude Enterprise, Claude Code, and the API. Highly agentic workload covers context-heavy, tool-heavy work, where cache reads make up most of the cost<\/p>\n<p><a href=\"https:\/\/claude.ai\/\" rel=\"nofollow noopener\" target=\"_blank\">Try Claude<\/a><a href=\"https:\/\/platform.claude.com\/\" rel=\"nofollow noopener\" target=\"_blank\">Start building<\/a><\/p>\n<p>1Terminal-Bench-Science 0.1: The standard error is \u00b13.5\u20134.5 pts per model. The public leaderboard (3 trials\/task, Claude Code harness) reports Claude Opus 5 at 30.0% and Claude Fable 5 at 21.4%; our setup reproduces them at 29.0% and 24.7%, respectively, both within noise<\/p>\n<p><sup>2<\/sup><strong>OSWorld 2.0:<\/strong> Scores are on the benchmark authors\u2019 August 2026 task release; Fable 5 and Opus 5 were re-run under the same conditions. Because the task files differ from earlier releases, these numbers aren&#8217;t directly comparable to previously published OSWorld 2.0 results, which is why no competitor score is shown<\/p>\n<p>3These three targets are (EGFR, Nipah G, 15-PGDH) and come from Adaptyv Bio\u2019s protein design competitions. The Nipah G comparison is against de novo designs targeting the receptor-binding site on the G head (best: ~8\u201312 nM, N1032). A de novo entry from Nick Boyd\/Escalante Bio that targets a different region (the stalk) reached ~1.4 nM (design_7), comparable to our best binder<\/p>\n<div style=\"clear:both;margin:30px 0 15px 0\">\n<p>\n    <strong>Related:<\/strong><br \/>\n    <a href=\"https:\/\/yoursite.com\/automation-training-benin\/\" title=\"Digital Automation Training Benin: 5 Winning Skills Employers Demand in 2026\" target=\"_blank\" rel=\"noopener\"><br \/>\n      Digital Automation Training Benin: 5 Winning Skills Employers Demand in 2026<br \/>\n    <\/a>\n  <\/p>\n<p>\n    &lt;a href=&quot;https:\/\/yoursite.com\/automation-africa\/&quot; title=&quot;WhatsApp <a href=\"https:\/\/justfineinfotech.com\/best-marketing-automation-software-2026-11-tools-compared-brevo\/\" title=\"Best Marketing Automation Software (2026): 11 Tools Compared | Brevo\">Marketing Automation<\/a> Africa: 6 Dangerous Mistakes Brands Make in Nigeria&#8221;&gt;<br \/>\n      WhatsApp Marketing Automation Africa: 6 Dangerous Mistakes Brands Make in Nigeria<br \/>\n    <\/a>\n  <\/p>\n<\/div>\n<div style=\"clear:both;margin:30px 0;padding:25px;background:#f8f9fc;border:1px solid #ddd;border-radius:8px;text-align:center\">\n<h3>Want to learn this practically?<\/h3>\n<p>Join <strong>Justfine Infotech<\/strong> and build real digital skills in AI, automation, web development, digital marketing, office productivity, e-commerce, freelancing and cybersecurity.<\/p>\n<p><strong>Available Programmes:<\/strong><br \/>\n  6 Weeks Certificate \u2022 3 Months Professional Certificate \u2022 6 Months Diploma \u2022 Full Professional Diploma<\/p>\n<p><strong>WhatsApp:<\/strong><br \/>\n  +229 01 57 57 99 15<br \/>\n  +229 01 66 68 11 60<\/p>\n<p><a href=\"https:\/\/api.whatsapp.com\/send?phone=2348132690270&amp;text=Hello\" target=\"_blank\" rel=\"noopener\">Enroll Now<\/a><\/p>\n<\/div>\n<p class=\"ani-source\">Source: <a href=\"https:\/\/www.anthropic.com\/claude-fable-and-mythos-5-1\" target=\"_blank\" rel=\"nofollow noopener\">www.anthropic.com<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Terminal-Bench-Science 0.1: The standard error is \u00b13.5\u20134.5 pts per model. The public leaderboard (3 trials\/task, Claude Code harness) reports Claude Opus 5 at 30.0% and Claude Fable 5 at 21.4%; our setup reproduces them at 29.0% and 24.7%, respectively, both within noise<\/p>","protected":false},"author":1,"featured_media":7254,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[63],"tags":[78,93,414,1617],"class_list":["post-7251","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-tools-chatgpt-updates","tag-claude","tag-fable","tag-introducing","tag-mythos"],"_links":{"self":[{"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/posts\/7251","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/comments?post=7251"}],"version-history":[{"count":1,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/posts\/7251\/revisions"}],"predecessor-version":[{"id":7253,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/posts\/7251\/revisions\/7253"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/media\/7254"}],"wp:attachment":[{"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/media?parent=7251"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/categories?post=7251"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/justfineinfotech.com\/fr\/wp-json\/wp\/v2\/tags?post=7251"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}