NVMe GPU Seedboxes
Download it, then stream it to your TV, phone or laptop. NVMe storage plus a shared Intel iGPU that transcodes video in Jellyfin, so each device gets a stream it can play.
- M1000 NVMe 150G
- 2.99€
monthly billing
- 150 GiB NVMe storage
- I/O: Archive (50)
- 1 Gbps networking
- 1 CPU core · Archive (50)
- 1 GB RAM
- 4 000 GiB traffic/mo
- No iGPU
- Token multiplier ×5
- Signup bonus 1M AI credits
- Sign Up!
Available: 30
- M1000 NVMe 400G
- 5.99€
monthly billing
- 400 GiB NVMe storage
- I/O: Standard (200)
- 1 Gbps networking
- 2 CPU cores · Standard (200)
- 2 GB RAM
- 10 000 GiB traffic/mo
- Shared Intel iGPU transcoding
- Token multiplier ×5
- Signup bonus 2M AI credits
- Sign Up!
Available: 30
- M1000 NVMe 800G
- 9.99€
monthly billing
- 800 GiB NVMe storage
- I/O: Standard (400)
- 1 Gbps networking
- 3 CPU cores · Standard (400)
- 6 GB RAM
- 20 000 GiB traffic/mo
- Shared Intel iGPU transcoding
- Token multiplier ×10
- Signup bonus 2M AI credits
- Sign Up!
Available: 30
- M1000 NVMe 1200G
- 19.99€
monthly billing
- 1 200 GiB NVMe storage
- I/O: Performance (700)
- 1 Gbps networking
- 4 CPU cores · Performance (700)
- 12 GB RAM
- 50 000 GiB traffic/mo
- Shared Intel iGPU transcoding
- Token multiplier ×20
- Signup bonus 5M AI credits
- Sign Up!
Available: 17
Every box runs in our own datacenters in Finland (EU).
Long Term Discounts: 5% 1 Year, 10% 2 Years, 15% 3 Years.
GPU Seedbox: stream to your TV, phone or laptop with hardware transcoding
A media file that plays fine on your computer often stutters or will not play at all on a TV app, a phone on mobile data or a hotel connection. The usual causes: the device cannot decode the format (HEVC, 10-bit, 4K HDR), or the connection is too slow for the original bitrate. A media server fixes this by transcoding: it converts the video on the fly into a format and bitrate the device can handle.
Transcoding in software eats CPU, and a few streams can be enough to stall it. On the 400G, 800G and 1200G plans the transcoding runs on the server's Intel iGPU (Intel Quick Sync) instead, which does the same work with a fraction of the CPU. Most servers in this group have Intel HD or UHD Graphics 630; some older ones have Intel HD Graphics 530.
- Formats: H.264 and HEVC sources are decoded in hardware and streamed out as H.264 that almost every device plays. On the HD/UHD 630 servers this includes 10-bit HEVC; the HD 530 servers decode 10-bit HEVC on the CPU, which is much slower.
- 4K HDR on non-HDR screens: on the HD/UHD 630 servers, HDR video can be tone mapped to SDR in hardware, so a 4K HDR file plays with normal colours on a regular TV, laptop or phone.
- Lower bitrate on the go: watch from your phone on mobile data or a slow hotel connection.
- Family and friends: several people can watch at the same time.
The iGPU is shared. Each server has one iGPU, shared by all users on that server, on a best-effort basis with no reserved capacity per account. As a guide, a server typically handles five or more 1080p transcodes at once, or (on the HD/UHD 630 servers) one to two 4K HDR to 1080p, in total across its users. Not supported: AV1 and Dolby Vision. The 150G plan has no iGPU access.
Getting started: Jellyfin installs with one command (install-media-stack.sh guide), then turn on hardware acceleration in the Jellyfin dashboard. See the Jellyfin article for the details. Plex is not available on shared seedboxes.
Why NVMe instead of SATA SSD
NVMe drives connect over PCIe instead of SATA: lower latency and many more requests in flight at once. A seedbox makes many small reads at the same time, so that is where it shows:
- Many torrents seeding at once: lots of peers reading different parts of different files at the same time.
- Library scans: Jellyfin, Sonarr and Radarr reading through thousands of files.
- Unpacking archives while torrents keep running.
- Seeking in a video while streaming starts playing from the new spot quickly.
Bandwidth is 1Gbps Full Duplex, shared among all users of the server. Traffic limit counts all upload (including HTTP/HTTPS) but only limits external (internet) upload speed of all protocols to 100Mbps - datacenter internal bandwidth remains unlimited and allows full server link speed to be utilized within servers in our datacenter. Minimum limit period is 3days. Traffic usage is calculated from past 30days - and is not set by calendar month or service renewal date, always rolling past 30 days.
RAM amount is the plan's per-account memory target and it is enforced: above it the server starts reclaiming your memory, and a somewhat higher hard ceiling allows short bursts. You will want to average around that amount of RAM usage to maintain top performance; especially with applications such as Deluge and Jellyfin.