Claude Fable 5.2 and Mythos 5.2: Anthropic's new flagship tops the AI Intelligence Index at $8 per million input tokens

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, upgrading the models it builds for the most daunting technical challenges. The two are the same underlying model: Fable 5.2 is the public version, with strong cybersecurity and biology safeguards, while Mythos 5.2 drops some of those safeguards for vetted organizations in Anthropic's Cyber and Life Sciences Verification Programs. Fable 5.2 is available today on the Claude Platform as claude-fable-5-2, on Amazon Web Services, Google Cloud and Microsoft Azure, and in Claude subscriptions. Anthropic says the new models communicate more clearly than Fable 5.1 and generate outputs up to 30% faster.

On the Artificial Analysis Intelligence Index, Claude Fable 5.2 at max effort scores 64.6, which as of November 9, 2026 ranks it #1 of the 59 model families Artificial Analysis scores, ahead of Google's Gemini 4 Carbon at 60.0 and Claude Opus 5.5 at 59.1. Its lower effort settings stay near the top: 62.2 at xhigh, 61.8 at high, 56.4 at medium and 53.9 at low. In Anthropic's own benchmark table, Fable 5.2 leads Opus 5.5 on FrontierCode 2.0 agentic coding (49.1% at xhigh against 45.5%), Humanity's Last Exam with tools (74.2% against 67.7%) and OSWorld 2.1 computer use (89.3% against 87.2%). It does not win everything: Gemini 4 Carbon beats it on AutomationBench business workflows (53.9% against 51.4%).

Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, down from Fable 5.1's $10 and $50, and an optional fast mode speeds up token generation by up to 2.5 times at $16 and $80. Running the full Artificial Analysis index costs $22,191 at max effort, $7,106 at high and $3,360 at medium, so the effort setting is the main cost control: Anthropic's docs set high as the API default, medium is the default in Claude apps and Claude Code, and adaptive thinking is always on. In Anthropic's testing, Fable 5.2 at high effort scored about level with Opus 5.5 at max effort on FrontierCode 2.0, for roughly half the cost per task.

Anthropic's prerelease stories are unusually concrete. A team of agents running an early Fable 5.2 migrated a critical production service from Python to Rust, and the service now starts about 26 times faster. One engineer had Fable 5.2 design a prototype operating system from an empty folder, coordinating over 1,000 subagents for eight days to produce a working OS that boots on a laptop. A thousand Fable 5.2 agents formalized the proof of Fermat's Last Theorem in 30 hours, where Opus 5.5 with the same setup completed 26% in the same time. Fable 5.2 also solved two open problems in black-hole physics, designed three working electronic devices with circuit-board maker Diode, and helped predict whole virus capsid structures for gene therapy.

For cybersecurity, Anthropic calls Mythos 5.2 its most capable model yet: it scored 89.6% on its internal CVE Rediscovery evaluation, finding 10.2% more vulnerabilities than Opus 5.5, and one security company said one in six confirmed bugs Mythos 5.2 found had been missed by every other frontier model it tested. Fable 5.2's safeguards produce about 55% fewer unnecessary blocks than Fable 5.1's, blocked queries fall back to Opus models, and Anthropic says its anti-distillation protections are its strongest ever, cutting the data harvested by distillers by 90% to 99%. In a dedicated sandbox-escape evaluation, Mythos 5.2 used planted credentials in 3.1% of scenarios, down from 39% for Mythos 5. External evaluators included the UK AI Security Institute and METR.