release_notes/v1p0.md

Release Notes

01/27/2025

• Stability and safety improvements

01/09/2025

• Support General Post-Training through NeMO

01/06/2025

• Initial release of Cosmos 1.0 along with the Cosmos paper
• 13 models were released in the Hugging Face as shown in the table below.
• Inference scripts for the models were released in the Cosmos repository.

Item	Model name	Description	Try it out
1	Cosmos-1.0-Diffusion-7B-Text2World	Text to visual world generation	Inference
2	Cosmos-1.0-Diffusion-14B-Text2World	Text to visual world generation	Inference
3	Cosmos-1.0-Diffusion-7B-Video2World	Video + Text based future visual world generation	Inference
4	Cosmos-1.0-Diffusion-14B-Video2World	Video + Text based future visual world generation	Inference
5	Cosmos-1.0-Autoregressive-4B	Future visual world generation	Inference
6	Cosmos-1.0-Autoregressive-12B	Future visual world generation	Inference
7	Cosmos-1.0-Autoregressive-5B-Video2World	Video + Text based future visual world generation	Inference
8	Cosmos-1.0-Autoregressive-13B-Video2World	Video + Text based future visual world generation	Inference
9	Cosmos-1.0-Tokenizer-CV8x8x8	Continuous video tokenizer with 8x8x8 compression ratio	Inference
10	Cosmos-1.0-Tokenizer-DV8x16x16	Discrete video tokenizer with 16x8x8 compression ratio	Inference
11	Cosmos-1.0-PromptUpsampler-12B-Text2World	Prompt upsampler for Text2World	Inference
12	Cosmos-1.0-Diffusion-7B-Decoder-DV8x16x16ToCV8x8x8	Diffusion decoder for enhancing Cosmos 1.0 autoregressive WFMs' outputs	Inference
13	Cosmos-1.0-Guardrail	Guardrail contains pre-Guard and post-Guard for safe use	Embedded in model inference scripts