SAVRN Model Hub · DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold
DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold GPU Requirements
DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold needs about 238 GB of GPU memory at 16-bit, 119 GB at 8-bit and 59.6 GB at 4-bit: its 99.4B parameters plus a 20% working margin. The cheapest setup at 16-bit is 1x MI325X from $2.00 an hour, about $1,460 a month around the clock. 4 of the 9 accelerators the SAVRN Index prices hold it on one card at 16-bit: B300, GB300 NVL72, MI325X, MI355X.
Every Accelerator, Every Precision
How many cards of each accelerator the SAVRN Index prices it takes to hold DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold, and what that many cards cost an hour at the lowest listed on-demand price. Memory needed: 238 GB at 16-bit, 119 GB at 8-bit, 59.6 GB at 4-bit.
| Accelerator | Memory per card | Lowest price per card | 16-bit | 8-bit | 4-bit |
|---|---|---|---|---|---|
| H100 Voltage Park |
80 GB | $1.99 | 3 cards $5.97/hr |
2 cards $3.98/hr |
1 card $1.99/hr |
| H200 GMI Cloud |
141 GB | $2.60 | 2 cards $5.20/hr |
1 card $2.60/hr |
1 card $2.60/hr |
| B200 Vultr |
180 GB | $3.50 | 2 cards $7.00/hr |
1 card $3.50/hr |
1 card $3.50/hr |
| GB200 NVL72 GMI Cloud |
186 GB | $8.00 | 2 cards $16.00/hr |
1 card $8.00/hr |
1 card $8.00/hr |
| MI300X Vultr |
192 GB | $1.85 | 2 cards $3.70/hr |
1 card $1.85/hr |
1 card $1.85/hr |
| MI325X Vultr |
256 GB | $2.00 | 1 card $2.00/hr |
1 card $2.00/hr |
1 card $2.00/hr |
| B300 Massed Compute |
268 GB | $6.60 | 1 card $6.60/hr |
1 card $6.60/hr |
1 card $6.60/hr |
| GB300 NVL72 Verda |
279 GB | $11.94 | 1 card $11.94/hr |
1 card $11.94/hr |
1 card $11.94/hr |
| MI355X Vultr |
288 GB | $2.59 | 1 card $2.59/hr |
1 card $2.59/hr |
1 card $2.59/hr |
Running It Around the Clock
| Precision | Cheapest setup | Per hour | Per month (730 hours) |
|---|---|---|---|
| 16-bit | 1x MI325X (Vultr) | $2.00 | $1,460 |
| 8-bit | 1x MI300X (Vultr) | $1.85 | $1,350 |
| 4-bit | 1x MI300X (Vultr) | $1.85 | $1,350 |
One copy of the model on rented cards, busy or idle. Serving more users at once takes more copies or more memory for their contexts.
Questions
How much VRAM does DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold need?
About 238 GB at 16-bit; about 119 GB at 8-bit; about 59.6 GB at 4-bit: the weights plus 20% for the runtime and a short context. A long context needs more.
What is the cheapest GPU setup to run DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold?
At 16-bit, 1x MI325X from $2.00 an hour, at the lowest on-demand price the SAVRN Index lists.
Can DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold run on a single H100?
Yes at 4-bit: an H100 has 80 GB and DeepSeek-V4.1-Flash-1017-EXL3-2.7bpw-TensorFold needs 59.6 GB at 4-bit.
Memory is the weights at that precision plus 20% for the runtime and a short context. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 24, 2026. Setups beyond eight cards, one server, are not listed.