Skip to content
The Decoder· Manuel Uth·Indexed by this site · Originally published· AI updates →· Its place in the scan record →

Anthropic's Claude Fable 5.2 does not generate images, but instead paints stroke by stroke with code

Anthropic's Claude Fable 5.2 doesn't generate images, it paints them

AI summary

Anthropic's Claude Fable 5.2 does not use diffusion models or external image models, but instead paints stroke by stroke with code. A photo-realistic parking garage shopping cart image used about 700,000 brushstrokes. Anthropic says it scored 64.9% on the internal RoboDojo Sim benchmark, higher than GPT-6 Astra's 29.8%, but researcher Aris Zhu's real robotic arm comparison shows the model's fine motor actions are still limited.

Read originalthe-decoder.com