Anthropic ships Fable 5.2 and Mythos 5.2 for hard technical work
Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2. The company describes them as models designed for the most ambitious open-ended technical work. Anthropic says the new models show improvements in coding, reasoning, and vision, generate outputs up to 30 percent faster than the 5.1 versions, and communicate more clearly with users.
Fable 5.2 and Mythos 5.2 share the same underlying model. Fable has strong cybersecurity and biology safeguards, which lets Anthropic make it available to all users. Mythos is the same model, but access goes through Anthropic's verification programs for cybersecurity and life sciences. Regular mode costs $8 per million input tokens and $40 per million output tokens. An optional fast mode costs $16 input and $80 output and speeds token generation by up to 2.5 times, according to Anthropic.
In one line: Anthropic has shipped Claude Fable 5.2 and Mythos 5.2, faster and cheaper specialist models that can run long agentic coding jobs and solve hard technical problems.
What is new in the 5.2 models
The headline change is performance on difficult work. Anthropic says Fable 5.2 and Mythos 5.2 are a step above Claude Opus 5.5 in almost all evaluated capabilities. On the FrontierCode 2.0 evaluation, Fable 5.2 at High effort scored about level with Opus 5.5 at Max effort for roughly half the cost per task. At Xhigh effort, it scored the highest of any released model tested.
Coding is the most concrete area. Over 90 percent of changes merged into Anthropic's core codebase are now written by Claude with human guidance. A team of agents running an earlier version of Fable 5.2 migrated a critical production service from Python to Rust. Anthropic says that service now starts about 26 times faster.
One Anthropic engineer asked Fable 5.2 to design a prototype operating system from an empty folder. The model coordinated over 1,000 subagents, worked around the clock for eight days, and produced a working operating system that boots on a laptop with its own core, drivers, storage, windows, login screen, and installer.
For teams shipping software with agentic tools, this is the kind of multi-step, long-running coordination that matters. If you are evaluating where specialist coding models fit, the coding and developer tools directory is a useful place to track what people actually use. The orchestration pattern, one strong model delegating to many subagents, also fits the tools we list under automation and agents.
Where Mythos 5.2 fits
Mythos 5.2 is the cybersecurity and life sciences twin. Anthropic says it is its most capable model for cybersecurity engineering. It catches more vulnerabilities than previous models and is better at reproducing and fixing them. A security company found that one in six confirmed bugs found by Claude Mythos 5.2 were missed by all of the other frontier models they tested. On an internal CVE Rediscovery evaluation, Mythos 5.2 found 10.2 percent more vulnerabilities than Claude Opus 5.5 and 14.4 percent more than Mythos 5.1. One tester found that previous models overstated the seriousness of about 1 in 20 findings, while Mythos 5.2 did not overstate any.
Science also shows up in the prerelease claims. Anthropic says Fable 5.2 solved two open problems in theoretical physics, both about black holes. It also says the model formalized the proof of Fermat's Last Theorem in 30 hours. Separately, TechCrunch reports that AI startup Physical Superintelligence announced a proof of the Aaronson-Ambainis conjecture using a non-public Anthropic model. Anthropic did not confirm whether Fable 5.2 was that model.
Mythos 5.2 is not generally available. Enterprises can use it through Anthropic's Cyber Verification Program or Life Sciences Verification Program. Claude Enterprise customers can also access Claude Security, which scans codebases for vulnerabilities using Mythos 5.2.
What changed for delivery and enterprise teams
Pricing and data controls changed too. TechCrunch notes that Anthropic lowered Fable's input price from $10 to $8 per million tokens. That makes Fable 5.2 cheaper than GPT-6 Astra, its primary competitor. Faster inference serving is available for double the price.
On AWS, Fable 5.2 is available on Amazon Bedrock through US Geo CRIS and Global CRIS inference profiles. Anthropic designates it as a Covered Model, so by default its use requires retention for up to 30 days and human safety review. The Enterprise Frontier Safeguards (EFS) system, built with AWS, combines zero data retention with misuse safeguards. Eligible customers can use Fable 5.2 with zero data retention through June 30, 2027. A version that keeps data in a customer-controlled bucket is expected but has not launched yet.
For teams on Claude Code, the open-source claude-api skill helps migrate to Fable 5.2. Running /claude-api migrate scans your working directory and updates model references to the Bedrock model ID, adjusts thinking settings, removes deprecated parameters, and suggests effort levels. The skill is also bundled into CodeRabbit, JetBrains IDEs including Junie, Resolve AI, and Warp.
What Kalpak would try
For client delivery, I would first test Fable 5.2 on the long-running agentic coding jobs where Opus 5.5 stalls or costs too much. The claim that High effort matches Opus 5.5 Max effort at half the cost per task is worth a side by side on real repositories. I would not switch daily driver models until that evaluation is done. The zero data retention option matters for regulated clients, and I would watch for the customer-controlled bucket version before promising it to anyone. This is the same discipline we describe in shipping client products without a dedicated dev team.
Quick answers
What is Claude Fable 5.2?
Claude Fable 5.2 is Anthropic's most capable generally available model for technical work. It improves coding, reasoning, and vision, runs agent teams, and costs $8 per million input tokens with faster output than the 5.1 version.
How is Mythos 5.2 different from Fable 5.2?
Mythos 5.2 uses the same underlying model but is available only through Anthropic's verification programs for cybersecurity and life sciences. Fable 5.2 is available to all users.
Can I use Fable 5.2 with zero data retention?
Yes, through Enterprise Frontier Safeguards on AWS. Eligible customers can use Fable 5.2 with zero data retention until June 30, 2027, while a customer-controlled bucket version is expected later.