MODEL
Claude Fable 5.2
Overview
Claude Fable 5.2 is Anthropic‘s 2026-11-09 frontier release, shipped alongside its sibling Claude Mythos 5.2 as the same underlying model with different guardrails — Fable 5.2 is the GA release with production safeguards, Mythos 5.2 is the restricted-access variant for vetted cyber and life-sciences organizations. API id claude-fable-5-2; pricing lands at $8 in / $40 out per Mtok standard, with Fast mode at $16/$80 and up to 2.5× throughput, and $0.20/Mtok cache reads — a ~20% price cut off Fable 5.1. Benchmarks vs Opus 5.5: FrontierSWE v2 69.9% vs 62.3%, Remote Labor Index 41.3% vs 21.3%, RoboDojo Sim 64.9% vs 39.4%, HLE-with-tools 74.2% vs 67.7%; FrontierCode 2.0 at 49.1% (Xhigh) is called “the highest of any released model tested”. Dario Amodei‘s letter frames the release around coding, agent-execution, and scientific-discovery gains — with two headline demos (a 1,000-agent run that formalized Fermat’s Last Theorem in 30 hours, and two open black-hole problems solved). Claude Code v2.1.321 ships claude-fable-5-2 as the default Fable model on 2026-11-09, the same day as the blog — zero-lag integration inside Anthropic’s own tool-chain.
Timeline
- 2026-11-10-AI-Digest — Anthropic ships Claude Fable 5.2 and Claude Mythos 5.2 on
2026-11-09— same underlying model with different guardrails, a pattern Anthropic has run since the Fable 5 / Claude Mythos 5 architecture in June2026, not new to today. Pricing$8/$40per Mtok standard (Fast$16/$80, up to2.5×throughput),$0.20/Mtok cache reads, API idclaude-fable-5-2. Benchmarks vs Opus 5.5: FrontierSWE v269.9%vs62.3%, Remote Labor Index41.3%vs21.3%, RoboDojo Sim64.9%vs39.4%, HLE-with-tools74.2%vs67.7%; FrontierCode 2.0 at49.1%(Xhigh) “highest of any released model tested”. Simon Willison confirmed primary-source integration viallm-anthropic 0.31the same day (max-effort pelican-on-a-bicycle ran out of tokens without returning an image). Claude Codev2.1.321setsclaude-fable-5-2as the default Fable model the same day. Reframe worth carrying: the “one weights set, two safety envelopes” mechanic is the original Fable / Mythos 5 architecture from June2026— CNBC’s “commercial-and-regulatory hedge” framing overlays an IPO-week angle on a tiering pattern that has been shipping for months. Carry as “frontier release at scheduled cadence with a~20%price cut off Fable5.1; the bifurcation is business-as-usual Anthropic, not an IPO-week invention”. Log against MOC - Major Companies and MOC - Open Source Models.
Key Developments
-
Scheduled Frontier Cadence, Not an IPO-Week Stunt (November 9, 2026): The capability gains over Opus 5.5 on agent-execution benchmarks (RLI
41.3%vs21.3%, RoboDojo64.9%vs39.4%) are the substance; CNBC’s “commercial-and-regulatory hedge” framing is an IPO-week overlay. Correct frame: routine release-cadence tier upgrade —~20%price cut off Fable5.1, standard capability lift, same Fable/Mythos architecture as before. -
Zero-Lag Substrate Ship in Claude Code v2.1.321: Claude Code
v2.1.321shipsclaude-fable-5-2as the default Fable model the same day the blog drops —1Mcontext window,$8/$40per Mtok,$0.20/Mtok cache reads. Paired with OpenSpecv1.17.0’s skills-migration deprecation on the same cycle, the integration-cadence story is zero-lag model adoption inside the first-party tool-chain — against competing coding tools, this is the actual advantage, not the benchmark deltas. -
Same Underlying Model as Mythos 5.2, Guardrail-Split Not Weight-Split: Fable 5.2 and Claude Mythos 5.2 are the same underlying model with different guardrails — Fable is the GA release with production safeguards, Mythos is the restricted-access sibling for vetted cyber and life-sciences organizations. Third instance of the sibling-SKU deployment pattern Claude Mythos 5 / Claude Fable 5 introduced in June
2026. -
Scientific-Discovery Framing — Two Named Demos: A
1,000-agent Fable 5.2 run formalized Fermat’s Last Theorem in30hours; two open black-hole problems solved. CNBC and Gizmodo picked these up as the “scientific-discovery” framing. FrontierCode 2.0 at49.1%(Xhigh) is positioned as “the highest of any released model tested” with the blog framing High-effort as “roughly matches Opus 5.5 at ~half the cost” rather than printing a head-to-head FrontierCode figure.
Related
See also: Anthropic, Claude Mythos 5.2, Claude Fable 5.1, Claude Mythos 5.1, Claude Fable 5, Claude Mythos 5, Claude Code, Claude Opus 5.5, Dario Amodei, Simon Willison, MOC - Major Companies, MOC - Open Source Models, MOC - Agentic Coding.