← The Polylog AI Intelligence Brief
Morning Edition · Thursday, November 12, 2026Published at 1:18 AM EST · New York
Claude Fable 5.2 raises Anthropic's FrontierCode score to 49.1% and cuts API prices by 20%
Mythos 5.2 uses the same underlying model with fewer cybersecurity and biology safeguards. It scores 89.6% on CVE rediscovery and is available only to verified organizations.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9. The two products share one underlying model. Mythos ships with fewer cybersecurity and biology safeguards and is offered only through Anthropic's Life Sciences and Cyber Verification programs, Implicator reported. Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, 20% below the $10 and $50 that Fable 5.1 charged. Anthropic says output is up to 30% faster, SD Times reported.
The headline numbers come from Anthropic's launch tables, as reproduced by DataCamp and Kiledjian. On FrontierCode 2.0, Fable 5.2 scores 49.1%, up from 32.4% for Fable 5.1 and above 45.5% for Claude Opus 5.5. Anthropic says Fable 5.2 at High effort roughly matches Opus 5.5 at Max effort for about half the cost per task. Mythos 5.2 reaches 89.6% on CVE (Common Vulnerabilities and Exposures) Rediscovery, up from 78.3% for Mythos 5.1. Officechai reports 76.3% on CyScenarioBench and 85.1% on a protein-binder design task.
The vendor tables also show results where Fable 5.2 does not lead. Gemini 4 Carbon beats it on AutomationBench, 53.9% to 51.4%. Opus 5.5 edges it on the Elo-rated AA-Briefcase, 1800 to 1785. On Terminal-Bench-Science, Fable 5.2 scores 74.6% and Mythos 5.2 scores 75.3%.
The one widely cited independent number comes from Artificial Analysis, which ranks Fable 5.2 first on its Intelligence Index with a score of 65. The same testing puts the cost of maximum effort at about $26.54 and 46.4 minutes per task. The cut in list price per token is therefore partly offset by heavy token use at high effort settings. Platform availability beyond the Claude API and Claude Platform on AWS was not confirmed in the announcement, ExplainX noted. Both models keep a 30-day retention requirement and are excluded from zero data retention unless Anthropic authorizes it.
What this means
Two things are happening at once. A coding gain of nearly 17 points on Anthropic's own benchmark keeps pressure on OpenAI and Google, whose GPT-6 Astra and Gemini 4 Carbon trail on most of the vendor's tables. A 20% price cut released a week before the IPO also shows Anthropic prioritizing token volume over margin. Until the system card's harness settings are checked, the FrontierCode jump should be treated as a vendor result. The single independent index agrees on the ranking but shows that the effective cost per task is high. The split between Fable and Mythos is the more durable change. Anthropic now sells the same model at two safeguard levels and restricts access to the less restricted version behind identity verification. That is a template for releasing capability with tiered access, and it connects to today's regional-policy story. The likelier outcome is that OpenAI and Google respond with price moves on their top coding tiers within weeks rather than with a new model. This depends on Fable 5.2 holding its lead in independent coding evaluations such as SWE-bench reruns. If those evaluations show the gain shrinking under neutral harnesses, the pressure to cut prices falls.
Part of a tracked trend
Frontier Labs Race on AI Coding Capability
Coding is becoming a primary competitive battleground among frontier labs, with incumbents standing up permanent coding teams and investing in new training stages (e.g. midtraining) to match leaders like Anthropic; expect recurring reorganizations, benchmarks, and model releases aimed specifically at code.
Townsquare
Start a discussion in Townsquare.
More from this edition
- Anthropic's Claude ban now reaches staff of China- and Russia-controlled companies wherever they work
- Anthropic markets 825 million shares at $70 to $78 for an initial public offering valued at up to $1.4 trillion
- AlphaGenome 2 finds 30% of causal variants on the TenK10K test, nearly double its predecessor
- Gemini Robotics 2.5 sorts metal parts with 96% success on Mercedes-Benz assembly lines
- Nvidia's 2-billion-parameter GR00T 2 learns chemistry-lab tasks from a handful of demonstrations and will ship open weights
- Google prices Gemini Decision 1 at $0.10 per million input tokens and charges nothing for output
- Five AI models chatted with 1,277 dating-app matches for six days, and only eight people asked whether they were talking to a bot
Comments
0No comments yet.