What GPU do I need to run nvidia/Cosmos-Reason2-8B?
8.8B parameters, published in BF16. View on Hugging FaceGated
Set up Cosmos-Reason2-8BCosmos-Reason2-8B is published by nvidia on Hugging Face, with 526,961 downloads and 214 likes to date. It's a Qwen3VLForConditionalGeneration model built for image-text-to-text, published natively in BF16, and gated: you'll need to accept the model's terms on Hugging Face before downloading weights.
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.
| Precision | Weight size | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| BF16 | 16.3 GB | 19.6 GB | RTX A5000 | 1 | $0.176/hr |
| FP8 (quantized) | 8.2 GB | 9.8 GB | RTX 4000 SFF Ada | 1 | $0.198/hr |
| INT4 (quantized) | 4.1 GB | 4.9 GB | RTX 3060 | 1 | $0.110/hr |
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run Cosmos-Reason2-8B at its published (BF16) precision: 1× RTX A5000, at $0.176/hr per GPU ($0.176/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Cosmos-Reason2-8B: KV cache by context length
The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.
KV-cache size for Cosmos-Reason2-8B is not published for this architecture, so no figure is shown.
Cosmos-Reason2-8B: common questions
Does Cosmos-Reason2-8B fit on a 24 GB GPU?
Yes. At BF16 it needs 19.6 GB of VRAM, so a 24 GB card holds it with 4.4 GB to spare. A 16 GB card is not enough for it at BF16.
Do I need approval to download Cosmos-Reason2-8B?
Yes. nvidia gates this repository on Hugging Face, so you have to accept its terms with a signed-in account before the weights will download. It does not change the 19.6 GB the model needs once you have them.
What is the least VRAM Cosmos-Reason2-8B can run in?
4.9 GB, at INT4 (quantized), which fits a 6 GB card, against 19.6 GB at BF16. That row assumes an INT4 checkpoint has actually been published for this model, so check its Hugging Face page before planning around it.
Does quantizing Cosmos-Reason2-8B lower the GPU bill?
Yes. At BF16 the cheapest live fit is one RTX A5000 at $0.176/hr. At INT4 (quantized) it drops to one RTX 3060 at $0.110/hr, provided a quantized checkpoint exists for it.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More Cosmos models
- Cosmos-Reason2-2B (2.4B, BF16)
- Cosmos-Reason1-7B (8.3B, BF16)
- Cosmos-H-Surgical (15.2B, BF16)
- Cosmos3-Super-Text2Image-4Step (64.0B, BF16)
- Cosmos3-Super-Image2Video (64.6B, BF16)
Lineage
Based on Qwen3-VL-8B-Instruct.
Alternatives at this size
Other models for image-text-to-text within about a third of Cosmos-Reason2-8B's 8.8B parameters, from other model lines.
- Qwen3.5-9B (9.7B, BF16)
- llava-1.5-7b-hf (7.1B, F16)
- gemma-3-12b-it (12.2B, BF16)
- UI-TARS-1.5-7B (8.3B, F32)
- Ovis2.5-9B (9.2B, BF16)
More on Cosmos-Reason2-8B
Fits on a 24 GB GPU at BF16: every model that fits in 24 GB.
Fits on a 12 GB GPU at FP8: every model that fits in 12 GB.
Fits on an 8 GB GPU at INT4: every model that fits in 8 GB.
Best vision-language models: how Cosmos-Reason2-8B ranks against the rest.
Related reading: RTX A5000 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, and Best GPU for LLM inference.
