A100 80GB SXM
A100 80GB SXM has 80 GB of memory and 2,039 GB/s of memory bandwidth. 409 open models in our directory fit at 4-bit.
NVIDIA lists 400 W for the standard SXM part; a custom-cooled HGX version can run up to 500 W.
Planning a multi-GPU deployment? Check NVLink domains and network topology as well as memory fit. A GPU specification alone does not establish the interconnect in a server or cloud allocation.
| Category | Fits | Tight | Too big | Total |
|---|---|---|---|---|
| LLM | 222 | 3 | 55 | 280 |
| Vision | 82 | 3 | 21 | 106 |
| Image | 24 | 0 | 0 | 24 |
| Video | 22 | 0 | 0 | 22 |
| Speech | 40 | 0 | 0 | 40 |
| Embedding | 19 | 0 | 0 | 19 |
Largest open models that fit at 4-bit
See all 409 in the tableA model fits when its 4-bit memory estimate (8k context) stays within 90% of usable memory, about 72.0 GB here. Showing the 30 largest of 409.
| Model | Type | Size | Memory at 4-bit |
|---|---|---|---|
| LLM | 103B | 69.5 GB | |
| Image | 83B | 56 GB | |
| LLM | 80.4B | 54.2 GB | |
| LLM | 78.1B | 51.7 GB | |
| Vision | 73.4B | 51.4 GB | |
| Vision | 72B | 51.3 GB | |
| Vision | 71.9B | 51.3 GB | |
| LLM | 70.6B | 50.3 GB | |
| LLM | 70.6B | 50.3 GB | |
| LLM | 70.6B | 49.5 GB | |
| LLM | 70.6B | 49.5 GB | |
| LLM | 41.8B | 45.3 GB | |
| LLM | 67.2B | 45.2 GB | |
| LLM | 51.6B | 36.8 GB | |
| LLM | 51.6B | 36.8 GB | |
| LLM | 51.6B | 35.2 GB | |
| LLM | 51.6B | 35.2 GB | |
| LLM | 49.1B | 34.7 GB | |
| LLM | 40.4B | 28.5 GB | |
| LLM | 40.4B | 28.5 GB | |
| LLM | 40.4B | 28.5 GB | |
| LLM | 40.4B | 28.5 GB | |
| LLM | 40.4B | 28.5 GB | |
| LLM | 36.2B | 26.2 GB | |
| LLM | 36.2B | 26.2 GB | |
| LLM | 35B | 24.9 GB | |
| Vision | 33.3B | 24.6 GB | |
| LLM | 31.2B | 24.1 GB | |
| LLM | 32.8B | 24 GB | |
| Speech | 33.7B | 24 GB |
These are memory checks, not speed or runtime compatibility results.
Compared with
All hardwareCapacity and bandwidth answer different questions: memory decides what fits, bandwidth drives how fast it runs.
Launch dates, useful-life evidence and ownership cost scenarios