Hosting provider immers·cloud has announced lower rental prices for its NVIDIA Tesla T4 GPU servers. The new rates start at just 16 ₽ (around $0.17) per hour. For anyone crunching AI models or transcoding video streams, this feels a bit like finding forgotten cash in a winter jacket pocket.
What’s Tesla T4?
Tesla T4 is a versatile GPU designed for real-time workloads. It blends efficiency with high performance and covers a wide range of use cases — from neural network inference to graphics rendering.
Key specs include:
-
2× Intel Xeon Gold 6240R CPUs (up to 4.0 GHz with Turbo Boost)
-
Support for Intel AVX-512 and DL Boost
-
Up to 3072 GB DDR4 ECC Reg memory (2933 MHz)
-
1920 GB SSD storage
-
16 GB GDDR6 VRAM
-
320 Gbit/s bandwidth
-
320 Tensor cores, 2560 CUDA cores
-
RT cores for graphics and ray tracing acceleration
Where It Shines
-
Neural network and LLM inference (up to 16 GB VRAM)
-
Video transcoding (H.264 / H.265, scaling, stream processing)
-
VDI and cloud-based desktops with GPU acceleration
-
Long-running sessions where energy efficiency matters most
Pricing and Discounts
immers·cloud is rolling out flexible billing options. Here’s an example with the teslat4-1.4.8.60 configuration:
| Billing Plan | Monthly Price (₽) | Monthly Price ($) | Discount |
|---|---|---|---|
| Pay-per-second | 14,349.60 | ≈ $155 | none |
| 30-day prepaid | 12,197.10 | ≈ $132 | 15% |
| 60-day prepaid | 21,524.40 | ≈ $233 | 25% |
With prepaid plans, the full amount is charged upfront, so account balance needs to cover the discounted cost.
Not Just T4
immers·cloud notes that the price drop isn’t limited to Tesla T4 servers. The same conditions apply to other GPU setups, including RTX 4090, V100, A100, and even H100.
In short, immers·cloud is making high-performance computing less of a luxury and more of an everyday tool.