This page was written and translated automatically by AI and checked by software against its sources, with no human editing. Specifications come from the sources listed; memory requirements and speeds are estimates calculated with published formulas. How we use AI
Cosmos-Reason2-32B has 32.8B parameters. At Q4_K_M with a context of 8,192 tokens, a single GPU needs 22.4 GiB of available memory for the result to be “Fits”.
File sizes are the real files on Hugging Face. The other columns give X, the available memory one GPU needs for the result to be “Fits” (largest buffer estimate ÷ 0.95 + 1 GiB), with llama.cpp, Flash Attention, F16 KV cache, 4 parallel sequences and micro-batch 512. They do not depend on any particular GPU.
Quantization
File size
X at 4,096 context
X at 8,192 context
X at 32,768 context
X at 131,072 context
Q6_K
25.04 GiB
28.1 GiB
29.2 GiB
35.5 GiB
61.0 GiB
Q5_K_M
21.62 GiB
24.6 GiB
25.7 GiB
32.0 GiB
57.5 GiB
Q4_K_M
18.40 GiB
21.3 GiB
22.4 GiB
28.7 GiB
54.2 GiB
Q4_0
17.42 GiB
20.3 GiB
21.3 GiB
27.7 GiB
53.2 GiB
IQ4_XS
16.48 GiB
19.3 GiB
20.4 GiB
26.7 GiB
52.2 GiB
Q3_K_M
14.87 GiB
17.7 GiB
18.8 GiB
25.1 GiB
50.6 GiB
IQ3_XXS
11.94 GiB
14.6 GiB
15.7 GiB
22.0 GiB
47.5 GiB
Q2_K
11.50 GiB
14.2 GiB
15.3 GiB
21.6 GiB
47.1 GiB
IQ2_M
10.58 GiB
13.2 GiB
14.2 GiB
20.6 GiB
46.1 GiB
IQ2_XXS
8.40 GiB
11.0 GiB
12.0 GiB
18.4 GiB
43.8 GiB
Smallest GPU for each quantization
For each quantization: the listed GPU with the least memory whose estimate is “Fits” on one card at 8,192 tokens of context. GPUs whose result is “Unknown” (no runtime profile, workstation and data-center cards, unified memory) are not considered.
“Fits” means llama.cpp's default --fit load check would place all layers on the GPU: our estimate of llama.cpp's buffers, plus its error margin, plus llama.cpp's 1 GiB per-card reserve, is within the memory counted as available. It does not promise that peak memory while generating fits. How results are decided
Context 8,192 tokens. Highest quantization that fits on one card, its longest context and the speed ceiling.
Unknown: memory configuration not established for this GPU; only “Doesn’t fit” would be shown.
Unknown
CUDA (NVIDIA): llama.cpp, CUDA back end (NVIDIA discrete GPU), all layers on the GPU, Flash Attention on, micro-batch (-ub) 512, F16 KV cache, context (-c) as shown; “Fits” is judged with llama-server’s default 4 parallel sequences (-np auto, unified KV cache) and “Doesn’t fit” with 1 sequence; idle GPU with no other programs using its memory.
ROCm on Linux (AMD): llama.cpp, ROCm back end on Linux (AMD discrete GPU), all layers on the GPU, Flash Attention on, micro-batch (-ub) 512, F16 KV cache, context (-c) as shown; “Fits” is judged with llama-server’s default 4 parallel sequences (-np auto, unified KV cache) and “Doesn’t fit” with 1 sequence; idle GPU with no other programs using its memory.
The speed ceiling is memory bandwidth divided by the bytes read per generated token. It is a theoretical upper bound; public llama.cpp benchmarks typically reach about half to four-fifths of it, and small models and mixture-of-experts models often less. Public benchmarks
Original weights
The model’s original weight files (safetensors, counted as described in the methodology) total 62.1 GiB. They are listed for reference only; every estimate on this page is for the GGUF files above, run with llama.cpp. nvidia/Cosmos-Reason2-32B · Methodology