--- title: "GPU specs for rent in Uzbekistan: early access | UZ Dynamics" description: "Early access by request: specs for the NVIDIA B300, B200, H200, H100 and other GPUs from NVIDIA’s official materials. Availability and form factor pending." url: "https://uzdynamics.ai/en/gpus" lang: "en" image: "https://uzdynamics.ai/en/opengraph-image/card?1237d0f845bd82b9" alternates: ru: "https://uzdynamics.ai/ru/gpus" uz: "https://uzdynamics.ai/uz/gpus" en: "https://uzdynamics.ai/en/gpus" --- > For an index of this site's pages, see [llms.txt](https://uzdynamics.ai/llms.txt). A markdown copy of each page is at its address with `.md` appended. Early-access requests # GPU rental: catalogue and specs Specs from NVIDIA’s official materials. Availability and form factor of each model are being confirmed. [Open the sizing tool](https://uzdynamics.ai/en/sizing) Memory is specified per GPU. Multiple GPUs have separate memory: using a larger model across them requires compatible software and model partitioning. NVLink does not make them one GPU. We don’t publish prices yet: terms and availability come with our reply. **GPU catalogue and specs** | GPU | Memory per GPU | Memory bandwidth | FP16/FP8, TFLOPS without sparsity | FP64, TFLOPS | GPU interconnect | Good for (sizing-tool estimate) | Actions | | --- | --- | --- | --- | --- | --- | --- | --- | | 96 GB and more | | | | | | | | | B300 Blackwell, SXM (being confirmed) 270 GB HBM3e LLMs up to 123B in FP8 on one GPU, multi-GPU training (NVLink) | 270 GB HBM3e | 7.7 TB/s | 2,250/4,500 | 1.2 | NVLink 1.8 TB/s | LLMs up to 123B in FP8 on one GPU, multi-GPU training (NVLink) | [Open details](https://uzdynamics.ai/en/gpus/b300) [Request](https://uzdynamics.ai/en/request?source=catalog%3Ab300&gpu=b300) | | B200 Blackwell, SXM (being confirmed) 180 GB HBM3e LLMs up to 123B in FP8 on one GPU, multi-GPU training (NVLink), FP64 computing | 180 GB HBM3e | 7.7 TB/s | 2,250/4,500 | 37 | NVLink 1.8 TB/s | LLMs up to 123B in FP8 on one GPU, multi-GPU training (NVLink), FP64 computing | [Open details](https://uzdynamics.ai/en/gpus/b200) [Request](https://uzdynamics.ai/en/request?source=catalog%3Ab200&gpu=b200) | | H200 Hopper, SXM (being confirmed) 141 GB HBM3e LLMs up to 70B in FP8 on one GPU, multi-GPU training (NVLink), FP64 computing | 141 GB HBM3e | 4.8 TB/s | 989.4/1,978.9 | 34; Tensor Core 67 | NVLink 900 GB/s | LLMs up to 70B in FP8 on one GPU, multi-GPU training (NVLink), FP64 computing | [Open details](https://uzdynamics.ai/en/gpus/h200) [Request](https://uzdynamics.ai/en/request?source=catalog%3Ah200&gpu=h200) | | RTX PRO 6000 Blackwell, Server Edition (being confirmed) 96 GB GDDR7 LLMs up to 70B in FP8 on one GPU, 3D rendering, video encoding | 96 GB GDDR7 | 1.6 TB/s | 467.8/935.6 | not listed | PCIe 5, no NVLink | LLMs up to 70B in FP8 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/rtx-pro-6000) [Request](https://uzdynamics.ai/en/request?source=catalog%3Artx-pro-6000&gpu=rtx-pro-6000) | | 80 GB | | | | | | | | | H100 Hopper, SXM (being confirmed) 80 GB HBM3 LLMs up to 32B in BF16 on one GPU, multi-GPU training (NVLink), FP64 computing | 80 GB HBM3 | 3.35 TB/s | 989.4/1,978.9 | 34; Tensor Core 67 | NVLink 900 GB/s | LLMs up to 32B in BF16 on one GPU, multi-GPU training (NVLink), FP64 computing | [Open details](https://uzdynamics.ai/en/gpus/h100) [Request](https://uzdynamics.ai/en/request?source=catalog%3Ah100&gpu=h100) | | A100 Ampere, SXM (being confirmed) 80 GB HBM2e LLMs up to 32B in BF16 on one GPU, multi-GPU training (NVLink), FP64 computing | 80 GB HBM2e | 2.04 TB/s | 312/none | 9.7; Tensor Core 19.5 | NVLink 600 GB/s | LLMs up to 32B in BF16 on one GPU, multi-GPU training (NVLink), FP64 computing | [Open details](https://uzdynamics.ai/en/gpus/a100) [Request](https://uzdynamics.ai/en/request?source=catalog%3Aa100&gpu=a100) | | 48 GB | | | | | | | | | L40S Ada Lovelace, PCIe (being confirmed) 48 GB GDDR6 LLMs up to 32B in FP8 on one GPU, 3D rendering, video encoding | 48 GB GDDR6 | 0.86 TB/s | 362.05/733 | not listed | PCIe 4, no NVLink | LLMs up to 32B in FP8 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/l40s) [Request](https://uzdynamics.ai/en/request?source=catalog%3Al40s&gpu=l40s) | | RTX 6000 Ada Ada Lovelace, Workstation (being confirmed) 48 GB GDDR6 LLMs up to 32B in FP8 on one GPU, 3D rendering, video encoding | 48 GB GDDR6 | 0.96 TB/s | 364/728.5 | not listed | PCIe 4, no NVLink | LLMs up to 32B in FP8 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/rtx-6000-ada) [Request](https://uzdynamics.ai/en/request?source=catalog%3Artx-6000-ada&gpu=rtx-6000-ada) | | RTX A6000 Ampere, Workstation (being confirmed) 48 GB GDDR6 LLMs up to 14B in BF16 on one GPU, 3D rendering, video encoding | 48 GB GDDR6 | 0.77 TB/s | 154.8/none | not listed | supports an NVLink bridge for 2 GPUs | LLMs up to 14B in BF16 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/rtx-a6000) [Request](https://uzdynamics.ai/en/request?source=catalog%3Artx-a6000&gpu=rtx-a6000) | | 24 GB | | | | | | | | | L4 Ada Lovelace, PCIe (being confirmed) 24 GB GDDR6 LLMs up to 14B in FP8 on one GPU, 3D rendering, video encoding | 24 GB GDDR6 | 0.3 TB/s | 121/242 | not listed | PCIe 4, no NVLink | LLMs up to 14B in FP8 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/l4) [Request](https://uzdynamics.ai/en/request?source=catalog%3Al4&gpu=l4) | | RTX A5000 Ampere, Workstation (being confirmed) 24 GB GDDR6 LLMs up to 8B in BF16 on one GPU, 3D rendering, video encoding | 24 GB GDDR6 | 0.77 TB/s | 111.1/none | not listed | supports an NVLink bridge for 2 GPUs | LLMs up to 8B in BF16 on one GPU, 3D rendering, video encoding | [Open details](https://uzdynamics.ai/en/gpus/rtx-a5000) [Request](https://uzdynamics.ai/en/request?source=catalog%3Artx-a5000&gpu=rtx-a5000) | † Usable memory per NVIDIA. The chip’s maximum memory is 288 GB. ‡ Usable memory per NVIDIA. The chip’s maximum memory is 192 GB. \* Dense FP16/BF16 and FP8 from the H100 SXM5 column of NVIDIA’s whitepaper: for the H200, NVIDIA lists only with-sparsity figures, the same as the H100 SXM’s. \*\* Derived: NVIDIA’s whitepaper per-SM Tensor Core rate × the 2,430 MHz boost clock in the product brief. For the Server Edition, NVIDIA gives FP16/FP8/FP4 only as rounded 1/2/4 PFLOPS peaks, the level of sparse figures; it publishes no dense ones. \*\*\* Derived: half the with-sparsity FP16 figure NVIDIA gives for the RTX A5000 (222.2 TFLOPS); NVIDIA gives no dense one. FP64 “not listed”: NVIDIA’s specs for these cards don’t give it; per NVIDIA’s whitepapers, FP64 on the GA102, AD102 and GB202 chips runs at 1/64 of the FP32 rate. Specs from NVIDIA’s official materials, checked 26 September 2026. [Open the sizing tool](https://uzdynamics.ai/en/sizing) ## GPU planning guides - [How to choose a GPU for LLM inference](https://uzdynamics.ai/en/guides/llm-inference) - [LLM VRAM: weights, KV cache and context](https://uzdynamics.ai/en/guides/vram-kv-cache) - [LoRA and QLoRA: plan your GPU configuration](https://uzdynamics.ai/en/guides/lora-qlora) - [Cloud GPU, dedicated server or API: which should you choose?](https://uzdynamics.ai/en/guides/gpu-cloud-vs-dedicated-vs-api) - [How to estimate GPU rental cost: compute, storage and idle time](https://uzdynamics.ai/en/guides/gpu-rental-cost) - [RTX 4090 or L40S: which GPU should you rent for AI?](https://uzdynamics.ai/en/guides/rtx-4090-vs-l40s) [All guides](https://uzdynamics.ai/en/guides) [Request early access](https://uzdynamics.ai/en/request?source=gpus) ```json {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"UZ Dynamics","item":"https://uzdynamics.ai/en"},{"@type":"ListItem","position":2,"name":"GPU catalogue","item":"https://uzdynamics.ai/en/gpus"}]} ```