# Anthropic Ships Claude Fable 5.2 With a 20% Price Cut and Keeps Its Cyber-Capable Mythos 5.2 Behind Vetting

Fable 5.2 scores 65 on the Artificial Analysis index, the top score among 74 models, but costs $26.54 per index task, and its counterpart Mythos 5.2 found 89.6% of known vulnerabilities on a rediscovery test.

- Published: 2026-11-11T06:08:31.146Z
- Canonical: https://polylog.news/ai/2026-11-11/anthropic-ships-claude-fable-5-2-with-a-20-price-cut-and-kee
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Anthropic News](https://www.anthropic.com/claude-fable-and-mythos-5-2), [VentureBeat](https://venturebeat.com/technology/anthropic-releases-claude-fable-5-2-and-mythos-5-2-pairing-new-record-high-benchmarks-with-20-api-cost-reduction), [The Decoder](https://the-decoder.com/claude-fable-5-2-and-mythos-5-2-cement-anthropics-ai-lead-just-before-its-ipo/), [Implicator](https://www.implicator.ai/anthropic-fable-5-2-mythos-5-2-price-cut/), [arXiv cs.AI](https://arxiv.org/abs/2611.12935)

Anthropic [released Claude Fable 5.2 and Claude Mythos 5.2](https://www.anthropic.com/claude-fable-and-mythos-5-2) on November 9 and describes them as its most capable models for coding, finance and the sciences. Most of the numbers come from Anthropic's own launch table. On FrontierCode 2.0, Fable 5.2 [scored 49.1%](https://venturebeat.com/technology/anthropic-releases-claude-fable-5-2-and-mythos-5-2-pairing-new-record-high-benchmarks-with-20-api-cost-reduction), against 45.5% for Claude Opus 5.5, 42.8% for OpenAI's GPT-6 Astra and 41.1% for Gemini 4 Carbon. Anthropic reports 69.9% on FrontierSWE v2, 89.3% on OSWorld 2.1 and 74.2% on Humanity's Last Exam with tools, where GPT-6 Astra scored 57.2%.

Independent testing confirms the ranking and adds a cost figure. Artificial Analysis [gave Fable 5.2 at maximum effort a 65](https://the-decoder.com/claude-fable-5-2-and-mythos-5-2-cement-anthropics-ai-lead-just-before-its-ipo/) on its Intelligence Index, first among 74 models listed. The same testing measured about 232 seconds to first answer token, 103 output tokens per second and $26.54 per index task. Fable is not ahead everywhere. On the Elo-rated AA-Briefcase knowledge-work test (a ranking system similar to chess ratings), Opus 5.5 leads 1800 to 1785. On AutomationBench, Gemini 4 Carbon leads 53.9% to 51.4%.

The API price fell [20% to $8 per million input tokens and $40 per million output](https://www.implicator.ai/anthropic-fable-5-2-mythos-5-2-price-cut/), from $10 and $50 for Fable 5.1. Even after the cut, Fable costs twice as much as Opus 5.5. Anthropic says output speed improved by up to 30%, and that claim has not been independently checked.

Mythos 5.2 matters more for governance. Anthropic describes it as the same underlying model as Fable 5.2 with certain cyber and biology safeguards removed. Access is [limited to vetted organizations](https://www.implicator.ai/anthropic-fable-5-2-mythos-5-2-price-cut/) in its Life Sciences and Cyber Verification Programs. On the CVE (Common Vulnerabilities and Exposures) Rediscovery benchmark, Mythos 5.2 [found 89.6% of known vulnerabilities](https://venturebeat.com/technology/anthropic-releases-claude-fable-5-2-and-mythos-5-2-pairing-new-record-high-benchmarks-with-20-api-cost-reduction), up from 78.3% for Mythos 5.1. Because both products share one set of weights, the access boundary depends on serving-time safeguards and customer vetting. The weights themselves do not differ.

New research addresses the monitoring layer that this kind of split relies on. A [paper posted to arXiv on November 11](https://arxiv.org/abs/2611.12935) defines topic monitoring as labeling the topic of every token in a response while the model generates it. The authors note that activation-based monitors running at generation time have so far been trained mainly to detect harmful content. Their method trains sparse autoencoder probes on prefill activations and applies them at decode time. If such probes scale, a lab could enforce finer access rules than a binary refusal.

## What this means

Anthropic now sells one model under two access regimes. Price and the safeguard layer separate the regimes, and the weights do not. Defensive security vendors and government cyber units that pass vetting gain a vulnerability-finding capability that is close to automated. Everyone else depends on classifiers holding up against jailbreaks. Leadership on vendor-reported coding scores is already expected from Anthropic. The news is the 20% price cut, the first concession on price at the top of its lineup, made days before its listing and one day before Google released a cheaper Carbon. The consensus describes Anthropic's lead as secure. The Artificial Analysis cost per task of $26.54 and the narrow CursorBench margin suggest the lead holds on difficult reasoning but not on the price of routine agentic work. The likelier outcome over the next one to two months is that enterprise coding spend moves to whichever model costs least for each completed task, with Fable kept for the hardest tasks. This depends on Carbon matching Fable on independent coding evals. The read is wrong if independent per-task cost data shows that Fable's higher accuracy makes it cheaper overall than Carbon on real repositories.

## What to watch

- Whether Google extends the $2/$10 rate past November 30. An extension would show that Google is buying market share deliberately, not running a launch promotion.
- Independent reproduction of the FrontierCode 2.0 and OSWorld 2.1 numbers. Most headline figures are Anthropic's own.
- Which organizations Anthropic admits to the Cyber Verification Program, especially any outside the United States. The list will show whether access to the model is becoming a tool of national security policy.