PulseAugur
EN
LIVE 10:58:24

Anthropic ships dual-model Claude Fable 5 with advanced coding and safety features

Anthropic has released Claude Fable 5, a model that the company deems too dangerous for unrestricted release. The model is a dual system: a public-facing version, Fable 5, uses a classifier to route potentially risky queries to the more capable but restricted Mythos 5 model, which is available only to vetted organizations. Fable 5 demonstrates significant advancements in coding benchmarks, outperforming previous models and showing impressive real-world application in tasks like large-scale code migration. However, access to Fable 5 now requires a 30-day data retention policy for all users, including enterprise clients, which Anthropic states is for safety monitoring rather than training. AI

IMPACT Sets a new SOTA for coding tasks, potentially accelerating enterprise adoption of AI for software development and engineering.

RANK_REASON New model release from a frontier lab (Anthropic) with detailed system description and capabilities. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on dev.to — Anthropic tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

COVERAGE [1]

  1. dev.to — Anthropic tag TIER_1 English(EN) · Peremptory ·

    Anthropic Ships a Model It Says Is Too Dangerous to Ship Without a Leash

    <p>Anthropic released Claude Fable 5 yesterday, and the product announcement itself is the most honest piece of AI marketing I've read in a while. The company released a model it considers, in its own framing, too dangerous to release without a leash, and then immediately release…