[sylvesterkaczmarek]

Summary

• clarify that the published four-step checkpoints are task-specific T2I and I2V students
• distinguish those checkpoints from the multi-task Cosmos3-Nano and Cosmos3-Super base models
• clarify that fixed-step sampling support does not imply every Cosmos3 checkpoint is distilled

Why

Issue #282 asked whether the presence of fixed-step sampling meant a distilled Cosmos3 variant was available. The project has since released Cosmos3-Super-Text2Image-4Step and Cosmos3-Super-Image2Video-4Step. A maintainer clarified that these are task-specific distilled models, while base multi-task Nano/Super distillation is separate work.

The distillation cookbook already lists the released students, but it does not explicitly state that scope. This small documentation change makes the distinction clear next to the distillation recipes.

Validation

Documentation-only change. Cross-checked against the published four-step model entries, the distillation cookbook, and the maintainer clarification in #282.

Closes #282

[sylvesterkaczmarek]

Refreshed this docs PR onto current main today. It is now 0 commits behind and mergeable, and the diff remains limited to the intended documentation file. This has been quiet for over two weeks. Could a maintainer please review/merge it, or let me know if any wording change is needed?

[sylvesterkaczmarek]

Could a maintainer review this clarification of the published four-step checkpoints? The validation workflow is still awaiting approval.