NVIDIA · Hopper · 80 GB · Qwen3.8 27B · Q4_K_M: ~130 tok/s
NVIDIA H100 80GB SXM5
The fastest single card you can run at home, if the word card stretches to a server module on a Chinese SXM5-to-PCIe adapter board: 80 GB of HBM3 at 3.35 TB/s. Expect to feed it up to 700 W and build the cooling around it.
Lowest price right now
$14,000
eBay · plus shipping · $175 per GB · observed 2026-10-18 19:13 UTC
plus an SXM5-to-PCIe adapter, about $900 to $1,000
Buy the cheapest on eBay, $14,000
plus shipping · Used · bestoec 110 · 100.0%
Price alert
One email per move, item plus shipping. Clear the price to hear about any drop. Manage alerts.
Price history
Tracking started today. The chart appears once there is a second day of data.
Where to buy
| Vendor | Price | Condition | Observed | |
|---|---|---|---|---|
| eBay cheapest active listing |
$14,000 plus shipping |
new and used | 2026-10-18 19:13 UTC | Go to eBay |
On eBay now
7 listings passed our filters at the last check. Cheapest delivered first. Seller reputation is feedback count and positive rate.
-
$14,000plus shipping -
$19,599.99free shipping -
$23,099free shipping -
$24,117.60free shipping -
$24,117.60free shipping -
$26,880free shipping
What it runs
For each model, grouped by vendor with the newest first: the largest quant that fits with 8K of context and the estimated speed for one user, at 8K and at 32K of context. On a card, a model too big for its memory can still run with part of the weights in system RAM; we assume 64 GB, and the speed shown accounts for it. The model you picked on the ladder is highlighted. Estimates from the spec sheet, not measurements: see how we estimate.
| Model | Quant | Needs at 8K | Fit | Max context | Decode at 8K | At 32K | Prefill tok/s |
|---|---|---|---|---|---|---|---|
| DeepSeek | |||||||
| DeepSeek V4.1 Flash 552B-A16B Sep 2026 | Q4_K_M | 338 GB | does not fit | ||||
| DeepSeek V4 Flash 284B-A13B Jul 2026 | IQ2_M | 98 GB | runs with 19 GB in system RAM | ~38 | ~36 | ~6,700 | |
| DeepSeek V3.1 671B-A37B Aug 2025 | Q4_K_M | 411 GB | does not fit | ||||
| Gemma | |||||||
| Gemma 4 12B May 2026 | Q8_0 | 14 GB | fits | 128K+ | ~170 | ~160 | ~14,000 |
| Gemma 4 31B Mar 2026 | Q8_0 | 35 GB | fits | 128K+ | ~66 | ~62 | ~5,600 |
| Gemma 4 26B-A4B Mar 2026 | Q8_0 | 28 GB | fits | 128K+ | ~180 | ~180 | ~46,000 |
| Gemma 3 27B Mar 2025 | Q8_0 | 31 GB | fits | 128K+ | ~73 | ~69 | ~6,300 |
| GLM | |||||||
| GLM-5.3 753B-A40B Aug 2026 | Q4_K_M | 462 GB | does not fit | ||||
| GLM-5.3 Flash 320B-A18B Aug 2026 | IQ2_M | 111 GB | runs with 31 GB in system RAM | ~21 | ~20 | ~4,800 | |
| GLM-4.6 355B-A32B Sep 2025 | IQ2_M | 125 GB | runs with 46 GB in system RAM | ~9.2 | ~7.5 | ~2,700 | |
| GLM-4.5 Air 106B-A12B Jul 2025 | Q5_K_M | 78 GB | fits | 16K | ~120 | ~14,000 | |
| gpt-oss | |||||||
| gpt-oss-120b Aug 2025 | Q4_K_M | 73 GB | fits | 96K | ~190 | ~170 | ~34,000 |
| gpt-oss-20b Aug 2025 | Q8_0 | 24 GB | fits | 128K+ | ~180 | ~170 | ~48,000 |
| Llama | |||||||
| Llama 4 Scout 109B-A17B Apr 2025 | Q4_K_M | 70 GB | fits | 59K | ~110 | ~89 | ~10,000 |
| Llama 3.3 70B Dec 2024 | Q8_0 | 79 GB | fits | 10K | ~28 | ~2,500 | |
| Llama 3.1 8B Jul 2024 | Q8_0 | 11 GB | fits | 128K+ | ~230 | ~170 | ~22,000 |
| MiMo | |||||||
| MiMo V2.6 Pro 1.02T-A42B Sep 2026 | Q4_K_M | 624 GB | does not fit | ||||
| MiMo V2.6 Flash 309B-A15B Sep 2026 | IQ2_M | 107 GB | runs with 27 GB in system RAM | ~27 | ~26 | ~5,800 | |
| Mistral | |||||||
| Mistral Medium 3.5 128B Mar 2026 | Q3_K_M | 67 GB | fits | 42K | ~33 | ~29 | ~1,400 |
| Mistral Small 4 119B-A6.5B Mar 2026 | Q4_K_M | 79 GB | fits | 9K | ~130 | ~27,000 | |
| Mistral Small 3.2 24B Jun 2025 | Q8_0 | 28 GB | fits | 128K+ | ~81 | ~71 | ~7,200 |
| Nemotron | |||||||
| Nemotron 3.5 Lightning 30B-A3B Aug 2026 | Q8_0 | 35 GB | fits | 128K+ | ~190 | ~190 | ~49,000 |
| Qwen | |||||||
| Qwen3.8 Flash-Next 125B-A6B Aug 2026 | Q4_K_M | 78 GB | fits | 89K | ~190 | ~180 | ~29,000 |
| Qwen3.8 27B Aug 2026 (your pick) | Q8_0 | 31 GB | fits | 128K+ | ~75 | ~71 | ~6,400 |
| Qwen3.6 27B Apr 2026 | Q8_0 | 31 GB | fits | 128K+ | ~75 | ~71 | ~6,400 |
| Qwen3.6 35B-A3B Apr 2026 | Q8_0 | 39 GB | fits | 128K+ | ~200 | ~190 | ~58,000 |
| Qwen3.5 9B Feb 2026 | Q8_0 | 11 GB | fits | 128K+ | ~220 | ~210 | ~19,000 |
| Qwen3 Coder 480B-A35B Jul 2025 | Q4_K_M | 296 GB | does not fit | ||||
| Qwen3 235B-A22B Jul 2025 | Q3_K_M | 118 GB | runs with 39 GB in system RAM | ~10 | ~9.2 | ~3,900 | |
| Qwen3 32B Apr 2025 | Q8_0 | 38 GB | fits | 128K+ | ~59 | ~50 | ~5,300 |
Specifications
- Memory
- 80 GB HBM3
- Memory bandwidth
- 3,350 GB/s
- Bus width
- 5120-bit
- Shader cores
- 16,896
- Tensor cores
- 528
- FP32
- 67.0 TFLOPS
- FP16 (dense)
- 989 TFLOPS
- Power
- 700 W
- Form factor
- SXM5 module: needs an SXM5-to-PCIe adapter board and serious cooling
- Interface
- PCIe 5.0 x16
- Architecture
- Hopper
- Launched
- 2022
Source: manufacturer specifications.