Anthropic Releases Claude Fable 5.2 and Claude Mythos 5.2…
Anthropic has released two new models, Claude Fable 5.2 and Claude Mythos 5.2, which share the same underlying architecture and differ in safety guardrail mechanisms and access permissions. Fable 5.2 scored 64.9% on the RoboDojo Sim robotics simulation benchmark, higher than the previous generation Fable 5.1's 27.6%, Opus 5.5's 39.4%, and GPT-6 Astra's 29.8%; on the BenchCAD 2.1 test it scored 83.5 points, up from 44.4 points for the previous generation. Anthropic also partnered with Diode to use the models to autonomously complete component selection, four-layer PCB routing, 3D enclosure CAD modeling, and software writing, designing three working devices. The two models' output speed is up to 30% faster than the previous generation and supports a fast mode of up to 2.5x speed. Fable 5.2 is now available to users and developers, while Mythos 5.2 is aimed at verified professional organizations.
Why read this
SelectedAI score 78
The original text provides the specific scores of the two new models on robotics control and 3D engineering design benchmarks, allowing comparison of capability changes against the previous generation and competitors.
Topics
Body · AI translation
English text is being prepared.
On November 9, Anthropic officially released two new models, Claude Fable 5.2 and Claude Mythos 5.2. Along with improvements in capabilities such as coding, reasoning, and visual understanding, the models further expand into application scenarios including robot control, 3D engineering design, and physical hardware development. According to the introduction, the two models use the same underlying architecture, with the main differences being the safety protection mechanisms and access permissions.
In terms of robot control, Claude Fable 5.2 demonstrated strong complex task execution capabilities. On the RoboDojo Sim robot simulation benchmark, the model achieved a score of 64.9%, higher than the previous generation Fable 5.1's 27.6%, Opus 5.5's 39.4%, and GPT-6 Astra's 29.8%. Anthropic also demonstrated the model controlling real robots to complete tasks such as manipulating a Rubik's cube, unscrewing bottle caps, and painting, reflecting the further extension of general large models into the field of physical world interaction.
In terms of 3D engineering design, Fable 5.2 scored 83.5 points on the BenchCAD 2.1 test, significantly higher than Fable 5.1's 44.4 points. This benchmark mainly evaluates the degree to which the geometric structures and circuit designs generated by the model match the reference solutions.
At the same time, Anthropic partnered with electronics design company Diode to conduct hardware development experiments. Based on natural language requirements, the model autonomously completed component selection, four-layer PCB circuit board routing, 3D enclosure CAD modeling, and software writing, ultimately designing three devices that work properly: a home voice assistant, a USB conference speaker, and a handheld electronic pet.
In terms of performance, the output generation speed of the two models is up to 30% faster than the previous generation, and they support a fast mode of up to 2.5x speed. Currently, Fable 5.2 has been opened to users and developers, while Mythos 5.2 is mainly aimed at verified professional institutions.
Submissions/Tips: tougao@youxituoluo.com
Article/Business Cooperation: Lin Nan (WeChat 19250561593)
Join the industry exchange group: Lin Nan (WeChat 19250561593)
Digital technology industry information and service platform
English text is being prepared.