Back to all offerings
NVIDIA A100 80GB
80 GB VRAM312 FP16 TFLOPS400 W TDP13 providers79 offerings13 regions
The NVIDIA A100 80GB is an enterprise-class GPU built on the Ampere architecture, featuring 80 GB of memory with 2,000 GB/s of memory bandwidth and a 400 W TDP. It delivers 19.5 TFLOPS of FP32 performance and 312 TFLOPS of FP16 tensor performance. Currently available from 13 providers starting at $0.33/GPU/hour, with a market median of $1.45/GPU/hour across 79 configurations.
Cheapest
$0.330/hr
Vast.ai
Median
$1.450/hr
across 79 offerings
Most expensive
$3.608/hr
Sesterce Cloud
90-day trend
-1.4%
median rate
Hardware specifications
Same across all providers
FP16 performance
312 TFLOPS
peak
TDP
400 W
Memory bandwidth
2,000 GB/s
Architecture
Ampere
FP32 performance
19.5 TFLOPS
Brand
NVIDIA
Series
A100
Market price history
Median price across all providers
Prices updated 1 hour ago
13 offerings from 13 providers
Sorted by price ascending. Compare configurations side-by-side.
| Provider | Count | vCPU | RAM | Region | Per GPU hour | Total/hr | Action |
|---|---|---|---|---|---|---|---|
| ×1 | 31 | 117 GB |  | $1.390 | $1.39 | Launch | |
| ×1 | 32 | 79 GB |   +8 | From$0.330 +1.9% 30d | From$0.33 | Launch | |
 Thunder ComputeReferral link | ×1 | 4 | 32 GB | -- | From$1.090 | From$1.09 | Launch |
| ×1 | 28 | 120 GB | | From$1.350 -1.8% 30d | From$1.35 | Launch | |
Up to$500credit | ×1 | 31 | 117 GB |  | From$1.390 -6.3% 30d | From$1.39 | Launch |
| ×4 | 124 | 480 GB | | From$1.418 -14.7% 30d | From$5.67 | Launch | |
 OblivusReferral link | ×1 | 28 | 120 GB | | From$1.470 0.0% 30d | From$1.47 | Launch |
| ×1 | 16 | 120 GB | | $1.490 0.0% 30d | From$1.49 | Launch | |
| ×1 | 28 | 120 GB | | $1.490 0.0% 30d | From$1.49 | Launch | |
 Sesterce CloudReferral link | ×1 | 16 | 160 GB |  | From$1.518 +10.3% 30d | From$1.52 | Launch |
| ×1 | 22 | 120 GB | -- | $1.790 | From$1.79 | Launch | |
| ×1 | 30 | 200 GB | | From$1.990 +20.1% 30d | From$1.99 | Launch | |
 Theta EdgeCloudTrending | ×1 | 10 | 80 GB | -- | $1.990 | From$1.99 | Launch |
Up to$1credit | ×8 | 64 | 512 GB | | From$2.010 +0.2% 30d | From$16.08 | Launch |
LLMs that fit on the NVIDIA A100 80GB
Estimated VRAM for popular open-weight LLMs at 16-bit (FP16/BF16), 8-bit (FP8/INT8) and 4-bit (INT4/MXFP4) precision, against 80 GB per GPU. Showing the 12 largest that fit.
| Model | Parameters | 16-bit | 8-bit | 4-bit |
|---|---|---|---|---|
| Kimi K2.5 Moonshot AI | 1T 32B active | — Not released | — Not released | 714 GB Tight on 8× · 7% spare |
| GLM-5.2 Z.ai | 753B | — Not released | 994 GB Too large | 579 GB 8× · 61 GB free |
| DeepSeek R1 DeepSeek | 671B 37B active | — Not released | 826 GB Too large | 481 GB 8× · 159 GB free |
| DeepSeek V4 Flash DeepSeek | 284B | — Not released | — Not released | 192 GB 4× · 128 GB free or 2× · 0% spare |
| Solar Open2 250B Upstage | 250B 15B active | 601 GB 8× · 39 GB free | 330 GB 8× · 310 GB free or 4× · 16% spare | 192 GB 4× · 128 GB free |
| Qwen3 235B-A22B Alibaba | 235B 22B active | 564 GB 8× · 76 GB free | 310 GB 4× · 9.7 GB free | 181 GB 4× · 139 GB free or 2× · 6% spare |
| Laguna-S 2.1 Poolside | 118B 8B active | 282 GB 4× · 38 GB free | 155 GB 2× · 4.8 GB free | 90 GB 2× · 70 GB free or 1× · 6% spare |
| gpt-oss-120b OpenAI | 117B 5.1B active | — Not released | — Not released | 78 GB 1× · 1.7 GB free |
| Qwen3.6 35B-A3B Alibaba | 35B 3B active | 86 GB 2× · 74 GB free or 1× · 11% spare | 47 GB 1× · 33 GB free | 28 GB 1× · 52 GB free |
| Qwen3 32B Alibaba | 32.8B | 79 GB 1× · 1.4 GB free | 43 GB 1× · 37 GB free | 25 GB 1× · 55 GB free |
| Gemma 4 31B Google | 31.3B | 75 GB 1× · 4.9 GB free | 41 GB 1× · 39 GB free | 24 GB 1× · 56 GB free |
| GLM-4.7 Flash Z.ai | 31.2B | 75 GB 1× · 5.1 GB free | 41 GB 1× · 39 GB free | 24 GB 1× · 56 GB free |
Estimates: model weights plus 20% for KV cache and runtime overhead, at short context lengths. Long contexts and large batches need more. Rows marked “tight” hold the weights but not that full margin. A ×N figure is the total VRAM across N of these GPUs and makes no claim about interconnect throughput. How we estimate this · Start from a model instead
Want more from this page?
See incorrect data?
Frequently asked questions
How much does the A100 80GB cost per hour?
The A100 80GB starts at $0.33 per GPU-hour, with a median of $1.45 across 79 available configurations. Actual pricing depends on the provider, region, and whether you rent on-demand, spot or reserved.
Which cloud providers offer the A100 80GB?
14 providers currently list the A100 80GB: Sesterce Cloud, Hyperstack, AceCloud, Cyfuture AI, Omega Gradient, Runcrate and 8 more. Compare their hourly pricing, regions and machine specs in the table above.
How much VRAM does the A100 80GB have?
The A100 80GB comes with 80 GB of VRAM, which determines the largest models and batch sizes it can hold in memory for training and inference.
Which LLMs can I run on the A100 80GB?
With 80 GB of VRAM, a single A100 80GB can serve models such as gpt-oss-120b, Qwen3.6 35B-A3B, Qwen3 32B, Gemma 4 31B. The table above lists the estimated VRAM for each model at 16-bit, 8-bit and 4-bit precision, including the multi-GPU configurations needed for larger models.
What's the biggest LLM I can run on multiple A100 80GB GPUs?
The largest models we track that fit across several A100 80GB cards are GLM-5.2 on 8 GPUs at 4-bit, DeepSeek R1 on 8 GPUs at 4-bit, DeepSeek V4 Flash on 4 GPUs at 4-bit. A multi-GPU configuration means you rent and pay for every card for the whole run, and splitting a model across cards costs throughput as well: the GPUs exchange activations on every token, so performance depends on how fast they are wired together. We do not track interconnect, so check the provider's node specification. The table above gives the smallest configuration for each model at 16-bit, 8-bit and 4-bit precision.
Where is the A100 80GB cheapest?
The lowest on-demand price we currently track for the A100 80GB is $0.33 per GPU-hour at Vast.ai. Because providers update pricing and availability frequently, check the live table above before you launch.