GPU types & models

GeForce GTX 1050 Ti 4 GB - entry-level graphics card for lightweight AI inference and basic content creation.

GeForce GTX 1060 6 GB - mid-range Pascal graphics card for 1080p gaming and lightweight AI experimentation.

NVIDIA Tesla V100 16 GB Volta - legacy datacenter accelerator still popular for CUDA development, scientific compute and AI research.

GeForce GTX 1080 8 GB - high-end Pascal graphics card for 1440p gaming and creative workloads.

GeForce GTX 1660 Super 6 GB - refreshed Turing card with faster GDDR6 memory for 1080p gaming.

NVIDIA GTX Titan X 12 GB - flagship Maxwell prosumer card for rendering, CUDA development and entry AI workloads.

GeForce RTX 3060 Laptop GPU 6 GB - mobile Ampere card for gaming notebooks and portable AI workloads. Distinct from the desktop RTX 3060 (12 GB).

GeForce RTX 3080 10 GB - high-end graphics card for 4K gaming, creative rendering and AI-assisted workflows.

GeForce RTX 4060 Ti 8 GB - efficient GPU for 1440p gaming, DLSS 3 frame generation and lightweight AI projects.

NVIDIA RTX A2000 12 GB - compact Ampere professional GPU for CAD, VR development and AI-assisted workflows.

NVIDIA Tesla P4 8 GB - low-power Pascal inference accelerator for video transcoding and edge AI workloads.

NVIDIA Quadro P4000 8 GB - single-slot Pascal workstation GPU for VR, rendering and creative production.

GeForce RTX 3060 12 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

GeForce GTX 1070 Ti 8 GB - higher-clocked Pascal model bridging the gap between the GTX 1070 and GTX 1080.

GeForce RTX 2070 8 GB - mid-range Turing card for 1440p ray-traced gaming and creator workloads.

GeForce RTX 3060 Ti 8 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

GeForce RTX 3070 8 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

GeForce RTX 3070 Laptop GPU 8 GB - mobile Ampere card for gaming notebooks and portable AI workloads. Distinct from the desktop RTX 3070 (8 GB, 220 W).

NVIDIA Titan Xp 12 GB - flagship Pascal prosumer card for content creation, rendering and entry AI workloads.

GeForce GTX 1080 Ti 11 GB - flagship Pascal card still capable for 4K gaming, rendering and entry AI workloads.

GeForce GTX 1660 Ti 6 GB - higher-clocked Turing model for 1080p gaming and light creator workloads.

GeForce RTX 2060 6 GB - entry-level Turing RTX card bringing real-time ray-tracing to mainstream gaming.

GeForce RTX 2080 Ti 11 GB - high-end graphics card for 4K gaming, creative rendering and AI-assisted workflows.

GeForce RTX 4060 8 GB - efficient GPU for 1440p gaming, DLSS 3 frame generation and lightweight AI projects.

GeForce RTX 2070 Super 8 GB - refreshed Turing model with more cores for creators and deep-learning hobbyists.

NVIDIA RTX A4000 16 GB - compact professional GPU for CAD, VR development and AI-assisted workflows.

NVIDIA Quadro RTX 4000 8 GB - single-slot Turing workstation GPU with Tensor Cores for CAD, rendering and AI inference.

GeForce RTX 3090 24 GB - powerhouse card for gaming, 3D rendering and AI research on a budget.

GeForce RTX 4070 Super 12 GB - refreshed Ada model with more cores for creators and deep-learning hobbyists.

NVIDIA Tesla P100 16 GB HBM2 - Pascal datacenter accelerator for HPC and early deep-learning training workloads.

GeForce RTX 4070 Ti 12 GB - higher-clocked Ada GPU ideal for 4K esports, Blender rendering and local generative-AI tasks.

GeForce RTX 4070 Ti Super 16 GB

GeForce RTX 5060 Ti 16 GB Blackwell - rumoured mid-range card targeting efficient 1440p gaming and entry AI workloads.

NVIDIA Titan V 12 GB HBM2 - Volta-based prosumer card with first-generation Tensor Cores for AI research and HPC.

NVIDIA Quadro RTX 5000 16 GB - Turing workstation GPU with Tensor Cores for VR, rendering and AI development.

GeForce RTX 3080 Ti 12 GB - enthusiast graphics card for 4K gaming, creative rendering and entry-level AI experimentation.

Rumoured GeForce RTX 5070 12 GB Blackwell - expected mid-range card targeting efficient 1440p gaming and entry AI workloads.

GeForce RTX 4060 Laptop GPU 8 GB - mobile Ada Lovelace card for gaming notebooks and portable AI workloads. Distinct from the desktop RTX 4060 (8 GB).

NVIDIA RTX A4500 20 GB - Ampere-based workstation card for real-time ray-tracing, simulation and AI inference.

NVIDIA Tesla P40 24 GB - Pascal inference accelerator with high VRAM, popular for self-hosted LLM inference on a budget.

NVIDIA Titan RTX 24 GB - flagship Turing prosumer card with Tensor Cores for AI research, 8K video editing and rendering.

GeForce RTX 4090 24 GB - ultimate consumer GPU for 8K gaming, neural rendering and AI model training.

GeForce RTX 5060 8 GB Blackwell - rumoured mid-range card targeting efficient 1440p gaming and entry AI workloads.

GeForce RTX 4080 Super 16 GB - refreshed Ada model with more cores for creators and deep-learning hobbyists.

NVIDIA RTX 6000 24 GB

GeForce RTX 3070 Ti 8 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

GeForce RTX 5060 Laptop GPU 8 GB - mobile Blackwell card with GDDR7 memory for gaming notebooks and portable AI workloads. Distinct from the desktop RTX 5060 (8 GB).

GeForce RTX 3090 Ti 24 GB - high-end graphics card for gaming, creative rendering and AI-assisted workflows.

Leaked GeForce RTX 5070 Ti 16 GB - projected sweet-spot Blackwell GPU for creators, gamers and small-scale machine-learning.

GeForce RTX 5080 16 GB Blackwell - anticipated high-end successor delivering faster ray-tracing and AI inference.

GeForce RTX 3050 8 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

GeForce RTX 4080 16 GB - high-end Ada Lovelace GPU delivering excellent 4K ray-tracing performance and generative-AI acceleration.

NVIDIA Tesla V100 32 GB Volta - legacy datacenter accelerator still popular for CUDA development, scientific compute and AI research. Shipped in both SXM2 and PCIe form.

The NVIDIA L4 Tensor Core GPU powered by the NVIDIA Ada Lovelace architecture delivers universal, energy-efficient acceleration for video, AI, visual computing, graphics, virtualization, and more.

NVIDIA T4 16 GB Tensor Core GPU (Turing) widely used for cost-efficient AI inference, video transcoding and lightweight training workloads.

GeForce RTX 4070 12 GB Ada - efficient GPU for 1440p gaming, DLSS 3 frame generation and lightweight AI projects.

NVIDIA RTX 4000 Ada 20 GB - single-slot Ada Lovelace GPU optimised for compact workstations, AI inferencing and real-time graphics.

NVIDIA RTX A5000 24 GB - versatile pro GPU for large-scene rendering, VR and deep-learning acceleration.

NVIDIA RTX PRO 4000 Blackwell 24 GB GDDR7 ECC - workstation GPU for professional content creation and AI workloads.

NVIDIA A10 24 GB.

NVIDIA CMP 170HX 8 GB HBM2e - passively-cooled Ampere GA100 mining card. Fuse- and firmware-locked to a PCIe 1.1 x4 link (~1 GB/s to the host) with heavily restricted general compute, so it is only useful for memory-bandwidth-bound work.

NVIDIA RTX 2000 Ada Generation 16 GB - compact professional GPU for CAD, VR development and AI-assisted workflows.

Intel Arc Pro B70 32 GB GDDR6 - Big Battlemage workstation and inference card, pairing 32 Xe2 cores and 256 XMX engines with 32 GB of memory for large-model inference, rendering and visualization.

NVIDIA Quadro RTX 8000 48 GB - ultimate AI and visual computing performance for the data center.

NVIDIA RTX PRO 5000 Blackwell 48 GB GDDR7 ECC - workstation GPU for advanced visualization, simulation and AI development.

NVIDIA RTX A6000 48 GB - flagship Ampere workstation GPU powering complex VFX, digital twins and AI pipelines.

NVIDIA RTX PRO 4500 Blackwell 32GB GDDR7 with error-correcting code (ECC)

NVIDIA A100 80 GB Tensor Core GPU - the proven workhorse for deep-learning training, data analytics and accelerated HPC.

GeForce RTX 5090 32 GB Blackwell flagship - tipped to exceed RTX 4090 in gaming, rendering and generative-AI.

NVIDIA Quadro GV100 32 GB HBM2 - flagship Volta workstation GPU with Tensor Cores for AI research, simulation and professional visualization.

NVIDIA A30 24 GB Ampere GPU designed for mainstream enterprise AI inference and mixed workloads.

GeForce GTX 1070 8 GB - mid-range graphics card for 1440p gaming, creative rendering and AI-assisted workflows.

NVIDIA A40 48 GB workstation GPU balancing real-time ray-tracing, AI acceleration and professional graphics rendering.

The NVIDIA RTX™ 4000 SFF Ada Generation packs a punch with the features, capabilities, and performance demanded by professionals—all in a compact GPU design.

GeForce RTX 4090 D 24 GB - China-market Ada Lovelace SKU with slightly reduced CUDA cores compared to the standard RTX 4090.

NVIDIA RTX 5000 Ada 32 GB - single-slot Ada Lovelace GPU optimised for compact workstations, AI inferencing and real-time graphics.

NVIDIA A800 80 GB PCIe - China-market Ampere accelerator with the same compute as A100 but reduced NVLink bandwidth.

NVIDIA L40S 48 GB Ada GPU, tuned for faster generative-AI inference and graphics workloads in datacenter environments.

NVIDIA A16 16 GB multi-instance GPU tailored for virtual desktops, AI inference and mixed graphics workloads in the cloud.

NVIDIA RTX PRO 6000 96 GB - universal AI and visual computing performance for the data center.

NVIDIA RTX PRO 6000 Blackwell Server Edition MIG 1g.24gb slice - a partitioned instance providing 24 GB of memory and ~1/4 of the full GPU compute for isolated AI inference and lightweight training workloads.

NVIDIA RTX 6000 Ada 48 GB - top-tier Ada workstation card delivering 91 TFLOPS for advanced visualization and AI training.

NVIDIA RTX 4500 Ada 24 GB - Ada Lovelace workstation GPU for advanced visualization, simulation and AI development.

GeForce RTX 2080 Super 8 GB - refreshed Turing flagship for 4K gaming, content creation and AI experimentation.

NVIDIA RTX 5880 Ada 48 GB - China-market Ada Lovelace workstation card with reduced CUDA cores compared to the RTX 6000 Ada.

NVIDIA L40 48 GB Ada Lovelace GPU delivering advanced ray-tracing and AI performance for enterprise visualization and generative content.

Intel Gaudi 2 AI accelerator - High performance acceleration for GenAI and LLMs.

NVIDIA RTX PRO 6000 Blackwell Server Edition MIG 2g.48gb slice - a partitioned instance providing 48 GB of memory and ~1/2 of the full GPU compute for isolated AI inference and medium training workloads.

NVIDIA A100 40 GB offers affordable access to Ampere Tensor Cores for research AI training, inference and scientific computing.

Flagship NVIDIA H100 Tensor Core GPU with 80 GB HBM3 for exascale HPC, large-language-model training and low-latency AI inference.

AMD Instinct MI300X 192 GB HBM3 - CDNA 3 accelerator targeting large-language-model training and high-performance computing.

NVIDIA H100 NVL dual-GPU solution offering 2x94 GB HBM3 for ultra-large LLM inference and memory-bound AI workloads.

NVIDIA RTX PRO 6000 Blackwell 96 GB GDDR7 with Confidential Computing enabled - hardware-isolated, encrypted execution for secure AI and visual computing in the data center.

AMD Instinct™ MI325X accelerators are designed to deliver leadership performance for Generative AI workloads and HPC applications.

NVIDIA GH200 96 GB - high-performance GPU for large-scale AI training and scientific computing.

NVIDIA H200 Tensor Core GPU with 141 GB HBM3e, optimized for generative AI, transformer models and data-intensive HPC simulations.

Built on the 4th Gen AMD CDNA™ architecture, AMD Instinct™ MI355X GPUs deliver leadership AI and HPC performance.

Dual-board NVIDIA H200 NVL (2x141 GB) accelerator designed for trillion-parameter model inference and massive memory bandwidth.

Next-generation NVIDIA Blackwell B200 with 180 GB HBM3e delivers breakthrough performance for frontier AI training and scientific computing.

Next-generation NVIDIA Blackwell B300 with 288 GB HBM3e delivers breakthrough performance for AI training and scientific computing.

NVIDIA Blackwell B200 180 GB HBM3e with Confidential Computing enabled - hardware-isolated, encrypted execution for frontier AI training and inference on shared infrastructure.

NVIDIA Blackwell B300 288 GB HBM3e with Confidential Computing enabled - hardware-isolated, encrypted execution for large-scale AI training and reasoning workloads.

NVIDIA GB300 Grace Blackwell Ultra Superchip pairs an Arm-based Grace CPU with a B300 GPU featuring 288 GB HBM3e, designed for trillion-parameter AI training and reasoning workloads.

NVIDIA A2 Tensor Core GPU - Entry-level GPU that brings NVIDIA AI to any server.

NVIDIA A2 8 GB vGPU profile - half of a 16 GB A2 Tensor Core GPU, shared between tenants for entry-level inference and virtualised graphics workloads.

NVIDIA A30 12 GB vGPU profile - half of a 24 GB A30 Ampere GPU, for mainstream enterprise inference that does not need a full card.