Anthropic’s official @claudeai account announced the two models at 18:57 UTC on 9 November, 48 minutes after AWS’s price file carried their prices, by our arithmetic. The post calls them “capable collaborators on ambitious projects in many fields”; the piece takes launch details, prices and benchmarks from Anthropic’s announcement and pricing page.
A thread of early experiments with Claude Fable 5.2:
A 3D model of a cell, built entirely from code one organelle at a time. Built by @kevin_t_ngo.
Anthropic’s thread of early Fable 5.2 experiments. The @claudeai account posted it at 15:20 UTC on 10 November, the day after launch; the opener shows a 3D model of a cell built in code, credited to builder Kevin Ngo (@kevin_t_ngo), and the thread closes with “Claude Fable 5.2 is now available. What will you explore?” It had about 85,000 views at 15:35 UTC. The thread adds no new availability step: Anthropic’s release notes already listed Fable 5.2 on the Claude API, AWS, Google Cloud and Microsoft Foundry at launch.
🚨 Claude Fable 5.2 and Mythos 5.2 are live in AWS official Bedrock price file
> file pushed today 18:09 UTC > both 5.2 models, 35 regions each > input $10 → $8 , output $50 → $40 > cache read $0.25 → $0.20 , every line exactly 20% cheaper than 5.1
Mythos 5.2 price is dated Oct 1 , Fable 5.2 is Nov 1
The X account @chetaslua flagged the price file at 18:35 UTC on 9 November, listing both 5.2 models, 35 regions and every line 20% cheaper than 5.1, and adding its own view that a launch is close. The piece checks each figure against AWS’s Price List API directly; the launch timing is the poster’s view, not something the file shows.
Fable 5.2 and Mythos 5.2 are now both listed in AWS Bedrock’s public prices catalogue
I thought we were going to get Fable 5.5 but ok this works for me too
LuminaBench, an AI news and benchmarks account, posted at 18:39 UTC on 9 November that Fable 5.2 and Mythos 5.2 were now both listed in AWS Bedrock’s public prices catalogue, four minutes after @chetaslua. Both posts came before Anthropic’s announcement appeared on its newsroom; the piece checks the listing against AWS’s Price List API directly.
Simon Willison posted at 05:42 UTC on 10 November that Anthropic’s updated prompting guide for Fable 5.2 carries a tip to make the model less likely to be cautious in how it improves existing code: tell it “This codebase is yours”. The post shows a screenshot of the guide’s section; the update box in this piece quotes the guide itself.
Anthropic launched Claude Fable 5.2 and Claude Mythos 5.2 on 9 November, its announcement says, priced at $8 per million input tokens and $40 per million output tokens. Its pricing page lists $10 and $50 for the 5.1 models, so the new pair costs 20% less on every token line, by our arithmetic. Anthropic says Fable 5.2 is available today on all platforms, including AWS, Google Cloud and Microsoft Azure.
The prices showed up first in Amazon Web Services’ official Bedrock price list, published at 18:09 UTC, which lists both models in 35 AWS regions at the same $8 and $40. Two X accounts, @chetaslua and LuminaBench, flagged the listing at 18:35 and 18:39 UTC, and Anthropic announced the models on its official @claudeai account at 18:57 UTC, 48 minutes after the file, by our arithmetic. Bloomberg and CNBC also reported the launch, which comes as Anthropic is reported to be preparing to market its IPO.
Business Pill · Usage-Based Pricing
An evergreen Business Engineer explainer on usage-based pricing. It teaches the general idea only and does not discuss Anthropic, AWS or Claude Fable.
The key insight: AWS’s official price list, published at 18:09 UTC, lists Claude Fable 5.2 and Mythos 5.2 in 35 regions at $8 input and $40 output per million tokens, 20% below the 5.1 models on every token line, by our arithmetic. Anthropic’s own pricing page lists the same figures, and the company says Fable 5.2 is available today on all platforms.
What Anthropic Announced
Anthropic calls the pair upgrades of its models for “the most daunting technical challenges”, and its post on X calls them “capable collaborators on ambitious projects in many fields”. It says “Claude Fable and Claude Mythos are the same underlying model”: Fable carries cybersecurity and biology safeguards and is open to all users, while Mythos 5.2 is offered through verification programs for cybersecurity and life-sciences work.
On Anthropic’s own benchmark table, Fable 5.2 scores 49.1% on FrontierCode 2.0 agentic coding at Xhigh effort, against 32.4% for Fable 5.1 and 45.5% for Opus 5.5. That is 16.7 points above 5.1, by our arithmetic. On the Remote Labor Index it scores 41.3%, 2.3 times Fable 5.1’s 17.9%, by our arithmetic. These are the company’s own results.
Anthropic says the models generate outputs up to 30% faster than the 5.1 versions. An optional fast mode, which it says speeds up token generation by up to 2.5 times, costs $16 per million input tokens and $80 per million output tokens. For most coding and knowledge tasks, Anthropic still recommends starting with Claude Opus 5.5.
Update, 9 Nov (Anthropic). The line circulating on X tonight comes from the coding section of Anthropic’s Fable 5.2 launch post: “Over 90% of the changes merged into Anthropic’s core codebase are now written by Claude (with human guidance).” The post does not say how changes are counted or over what period.
The same section gives two examples. A team of agents running an earlier version of Fable 5.2 “migrated a critical production service from Python to Rust,” which Anthropic says now starts about 26 times faster. And one engineer gave the model an empty folder and the task of designing a prototype operating system: it “coordinated over 1,000 subagents, working around the clock for eight days,” producing a system that boots on a laptop. For an outside view, Node.js creator Ryan Dahl wrote on X earlier today that “software engineering has transitioned to 100% AI written.”
Update (10 Nov): Anthropic’s prompting guide for Fable 5.2. Anthropic’s prompting guide for Claude Fable 5.2 tells developers how the new model behaves differently from earlier ones. Under the heading “Say who owns the code”, it says: “Claude Fable 5.2 responds closely to what the prompt says about who owns the code it works on. In Anthropic’s testing on small coding tasks, when a prompt said nothing about ownership or presented the code base as someone else’s, the model tended to add to the existing structure, even when that repeated existing code. It also tended to report nearby problems instead of fixing them.”
The guide’s suggested sentence is: “This codebase is yours: what it’s for and how it’s built are your call, and you’ll keep building on it.” With it, the guide says, “the model merged repeated code in about half of the tries with this sentence, and in under a tenth of the tries without it”. The cost is stated too: “the model also used about 1.5 times the output tokens with the sentence and made several changes that nobody had asked for, many of them new tests and documentation.” The guide adds: “review the diff and not only the summary.”
Two other changes in the guide carry a token cost. On turns that end before the work is done, it offers system-prompt text and says that, in replayed sessions at medium effort, “about half as many sessions ended with doable work left” while “Output tokens rose by about 30 percent.” On effort, it says that “At high and above, the model thinks before almost every reply, including a reply to a greeting”, and that an instruction to answer directly “didn’t stop the model from thinking at medium and above.” Simon Willison’s post on the tip is embedded above.
What AWS’s Price File Showed First
The record sits in the AWS Price List API under the AmazonBedrockFoundationModels offer. Its newest version, 20261109180952, carries a publication time of 18:09:52 UTC on 9 November. The product names read “Claude Fable 5.2 (Amazon Bedrock Edition)” and “Claude Mythos 5.2 (Amazon Bedrock Edition)”, next to the existing 5.1, 5 and Mythos Preview entries.
The previous version, published in September, lists Fable 5, Fable 5.1, Mythos 5, Mythos 5.1 and Mythos Preview, and no 5.2 model. The price terms are dated 1 November for Fable 5.2 and 1 October for Mythos 5.2; the file does not say why the two dates differ.
Amazon Bedrock on-demand prices per million tokens, global cross-region tier, us-east-1, from the AWS Price List API version published 9 November 2026. The 20% change on each line is our arithmetic; Mythos 5.2 lists the same prices as Fable 5.2.
Every Line Is 20% Lower
In the us-east-1 file, the global cross-region prices per million tokens move as follows. Input falls from $10 to $8, output from $50 to $40, and five-minute cache writes from $12.50 to $10. One-hour cache writes fall from $20 to $16, and cache reads from $0.25 to $0.20. Each is a 20% cut, by our arithmetic, and Mythos 5.2 lists exactly the same prices as Fable 5.2.
In-region standard prices follow the same ratio: $8.80 input and $44 output for the 5.2 models, against $11 and $55 for 5.1 in us-east-1. The two AWS GovCloud (US) regions list $9.60 input. A job that reads one million tokens and writes one million costs $48 on 5.2 at global prices, against $60 on 5.1, by our arithmetic. Output still costs five times input on both versions.
Update (10 Nov): Scale Labs’ leaderboard puts Fable 5.2 first on the Remote Labor Index.
Scale Labs, which runs the Remote Labor Index with the Center for AI Safety, said on X that Fable 5.2 “successfully automates 41.3% of projects, nearly doubling the performance of the next-best model Opus 5.5”. That matches the 41.3% in Anthropic’s own results above.
The Decoder reported, citing the leaderboard, that the next places go to Opus 5.5 at 21.25% and OpenAI’s GPT 6 Astra at 20.83%, with Google’s Gemini Argon 4 fifth at 17.08%. Fable 5.2’s 41.3% is 1.94 times Opus 5.5’s score, by our arithmetic, and 16.5 times the 2.5% the top agent, Manus, scored when the benchmark launched, as the leaderboard page says.
The index scores whether an agent’s work on real freelance projects would be accepted by a paying client, judged by human reviewers, per The Decoder. Scale Labs said Fable 5.2’s biggest gains came in editing audio and music, designing architecture and interior spaces, and creating 3D models, The Decoder reported.
The X account @chetaslua flagged the file on X at 18:35 UTC, 26 minutes after it was published. On 1 November, a separate screenshot showed a Mythos 5.5 name in Anthropic site code.
The Structural Read
As we read it, the cut is uniform: input, output, cache writes and cache reads all fall by the same 20%, so the ratio between them does not change and output still costs five times input.
AWS Price List API, disclaimer field of the Bedrock foundation-model price file, 9 November 2026
“This pricing list is for informational purposes only”
Three Implications
FOR BEDROCK BUYERS Global input is $8 per million tokens for both 5.2 models in every region listing them; the two GovCloud (US) regions list $9.60 in-region input.
FOR AGENT WORKLOADS Cache reads fall from $0.25 to $0.20 per million tokens on the global tier, the line our Fable 5.1 coverage tied to the cost of agent loops.
FOR ANTHROPIC’S OWN API Anthropic’s own pricing page lists the same $8 and $40 for Fable 5.2 and Mythos 5.2, against $10 and $50 for the 5.1 models.
The framework argues: “an agent is a process that consumes tokens to produce work. The new job of finance is to manage the ratio between the two”.
As we read it, a 20% lower price per token improves that ratio only if the work per token holds; AWS’s file prices the tokens and says nothing about the work. The framework predates this listing and does not discuss Fable 5.2.
What Is Not Established
The benchmark figures are Anthropic’s own and have not been independently reproduced. The AWS file does not say whether Mythos 5.2 needs the same verification-program access on Bedrock as on Anthropic’s platform, and the file’s different term dates, 1 October for Mythos 5.2 and 1 November for Fable 5.2, are not explained.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Anthropic launched Claude Fable 5.2 and Mythos 5.2 at $8 input and $40 output per million tokens, 20% below the 5.1 models on every line, by our arithmetic. AWS’s price file carried the same prices in 35 regions 48 minutes before Anthropic’s post on X, by our arithmetic.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. Launch details, benchmarks and prices come from Anthropic’s announcement and pricing page of 9 November. Bedrock model names, prices, dates and the region count come from AWS’s public Price List API (AmazonBedrockFoundationModels, version 20261109180952), read directly. The percentage changes, point gaps and ratios are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
