NVIDIA L4
৳147*/hrAda24 GBPCIe
Fast inference, video AI, efficient fine-tuning
GPU cloud in Bangladesh
Access cloud GPU in Bangladesh B200,
H100, A100 and L4 by the hour.
AI COMPUTER CLOUD BANGLADESH
ComputeBD offers Bangladesh-focused AI computing through invite-based early access, with Jupyter workflows, BDT-first pricing and multiple accelerator classes.
WHY COMPUTEBD
Compute when needed instead of buying for the peak
High-end accelerators are expensive and evolve quickly. Researchers, startups and teams may not need to own hardware sized for their peak workload.
ComputeBD aims to make AI Computer time the clear unit: choose an accelerator, run the workload, see usage and stop when the job is finished.
NVIDIA A10
৳196*/hrAmpere24 GBPCIe
Rendering, vision, medium training workloads
NVIDIA L40S
৳343*/hrAda48 GBPCIe
Generative AI, rendering, larger fine-tunes
NVIDIA A100
৳638*/hrAmpere80 GBSXM
LLM fine-tuning, research, high-throughput training
NVIDIA H100
৳1,047*/hrHopper80 GBSXM
Large model training, transformer workloads, HPC
NVIDIA H200
৳1,390*/hrHopper141 GBSXM
Memory-heavy LLMs, retrieval, frontier research
Common questions
FAQ
Is there an AI cloud or machine-learning cloud in Bangladesh?
Yes. ComputeBD is a GPU cloud built for users in Bangladesh: NVIDIA GPUs for machine learning, deep learning, LLM training and inference, with Jupyter in the browser and prices in taka.
Which GPU should I use for deep learning?
An L4 for small models and experiments, an L40S for image models and medium training, an A100 or H100 for large language models. Workload-by-workload advice is in the GPU workload map.
Is it like having my own GPU server?
For your work, yes: the whole GPU is in your session and you can install packages and run any code. The difference is that you do not buy, host or maintain a server — you run it by the hour and stop it.
How is it different from the government GPU cloud?
The government shared GPU facility launched in January 2026 is mainly for universities and research, by application. ComputeBD is self-service for anyone — students, freelancers, startups — paid in taka, whenever you need it.