Anthropic releases new Mythos and Fable models amid RSI concerns

Anthropic today unveiled the next generation of its flagship models, Mythos 5.2 and Fable 5.2. Both versions claim stronger performance in coding, reasoning and vision, and deliver clearer communication with end users.

Fable 5.2 reduces the price per million input tokens from $10 to $8, making it cheaper than the competing GPT‑6 Astra, and cuts inference latency by roughly 30 % across all effort levels. Faster serving is also offered at double the price, and the model now supports zero‑data‑retention policies through Anthropic’s Enterprise Frontier Safeguards, addressing a key enterprise requirement.

Why it matters for GPU / AI infrastructure

For GPU cloud providers and AI infrastructure planners, the lower cost and faster inference of Fable 5.2 translate into higher throughput per GPU and more efficient resource allocation, while ZDR compliance eases integration into regulated environments. These advances help accelerate enterprise AI workloads and support the scaling strategies of providers such as AiGPU.