Text2026-06-25
Qwen3-Max raises the open-weight bar
Alibaba's Qwen3-Max ships open weights with strong coding and reasoning, narrowing the gap to closed frontier models.
Source: qwenlm.github.io
Best consumer GPU prices
The cheapest current listing for each gaming and prosumer GPU. See the full price comparison for everything.
| GPU | Manufacturer | Architecture | VRAM | Best | Vendor | Buy |
|---|---|---|---|---|---|---|
| GeForce RTX 5090 | NVIDIA | Blackwell | 32 GB GDDR7 | $3,405 | PLE Computers | Buy ↗ |
| GeForce RTX 4090 | NVIDIA | Ada Lovelace | 24 GB GDDR6X | $2,940 | PLE Computers | Buy ↗ |
| GeForce RTX 5080 | NVIDIA | Blackwell | 16 GB GDDR7 | $1,686 | PLE Computers | Buy ↗ |
| GeForce RTX 4080 SUPER | NVIDIA | Ada Lovelace | 16 GB GDDR6X | $1,547 | PLE Computers | Buy ↗ |
| Radeon RX 7900 XTX | AMD | RDNA 3 | 24 GB GDDR6 | $1,392 | PLE Computers | Buy ↗ |
| GeForce RTX 3090 | NVIDIA | Ampere | 24 GB GDDR6X | $1,315 | PLE Computers | Buy ↗ |
| GeForce RTX 5070 Ti | NVIDIA | Blackwell | 16 GB GDDR7 | $1,237 | PLE Computers | Buy ↗ |
| GeForce RTX 4070 Ti SUPER | NVIDIA | Ada Lovelace | 16 GB GDDR6X | $1,206 | PLE Computers | Buy ↗ |
| Radeon RX 7900 XT | AMD | RDNA 3 | 20 GB GDDR6 | $1,036 | PLE Computers | Buy ↗ |
| Radeon RX 9070 XT | AMD | RDNA 4 | 16 GB GDDR6 | $1,005 | PLE Computers | Buy ↗ |
| GeForce RTX 4070 SUPER | NVIDIA | Ada Lovelace | 12 GB GDDR6X | $912 | PLE Computers | Buy ↗ |
| Radeon RX 9070 | AMD | RDNA 4 | 16 GB GDDR6 | $896 | PLE Computers | Buy ↗ |
| GeForce RTX 5070 | NVIDIA | Blackwell | 12 GB GDDR7 | $896 | PLE Computers | Buy ↗ |
| Radeon RX 6950 XT | AMD | RDNA 2 | 16 GB GDDR6 | $818 | PLE Computers | Buy ↗ |
| GeForce RTX 3080 | NVIDIA | Ampere | 10 GB GDDR6X | $757 | PLE Computers | Buy ↗ |
| Radeon RX 7800 XT | AMD | RDNA 3 | 16 GB GDDR6 | $741 | PLE Computers | Buy ↗ |
| GeForce RTX 5060 Ti 16GB | NVIDIA | Blackwell | 16 GB GDDR7 | $695 | PLE Computers | Buy ↗ |
| Radeon RX 6800 XT | AMD | RDNA 2 | 16 GB GDDR6 | $649 | PLE Computers | Buy ↗ |
| Radeon RX 7700 XT | AMD | RDNA 3 | 12 GB GDDR6 | $634 | PLE Computers | Buy ↗ |
| GeForce RTX 4060 Ti 8GB | NVIDIA | Ada Lovelace | 8 GB GDDR6 | $586 | PLE Computers | Buy ↗ |
| Radeon RX 9060 XT 16GB | AMD | RDNA 4 | 16 GB GDDR6 | $572 | PLE Computers | Buy ↗ |
| GeForce RTX 3070 | NVIDIA | Ampere | 8 GB GDDR6 | $509 | PLE Computers | Buy ↗ |
Best budget consumer GPU prices
| GPU | Manufacturer | Architecture | VRAM | Best | Vendor | Buy |
|---|---|---|---|---|---|---|
| Radeon RX 6700 XT | AMD | RDNA 2 | 12 GB GDDR6 | $463 | PLE Computers | Buy ↗ |
| GeForce RTX 4060 | NVIDIA | Ada Lovelace | 8 GB GDDR6 | $447 | PLE Computers | Buy ↗ |
| Arc B580 | Intel | Battlemage | 12 GB GDDR6 | $417 | PLE Computers | Buy ↗ |
| Arc A770 16GB | Intel | Alchemist | 16 GB GDDR6 | $401 | PLE Computers | Buy ↗ |
| GeForce RTX 3060 | NVIDIA | Ampere | 12 GB GDDR6 | $401 | PLE Computers | Buy ↗ |
| Radeon RX 7600 | AMD | RDNA 3 | 8 GB GDDR6 | $370 | PLE Computers | Buy ↗ |
| GeForce GTX 1080 Ti | NVIDIA | Pascal | 11 GB GDDR5X | $370 | PLE Computers | Buy ↗ |
| Arc B570 | Intel | Battlemage | 10 GB GDDR6 | $354 | PLE Computers | Buy ↗ |
| GeForce RTX 2060 SUPER | NVIDIA | Turing | 8 GB GDDR6 | $324 | PLE Computers | Buy ↗ |
| Radeon RX 6600 | AMD | RDNA 2 | 8 GB GDDR6 | $292 | PLE Computers | Buy ↗ |
| Arc A750 | Intel | Alchemist | 8 GB GDDR6 | $292 | PLE Computers | Buy ↗ |
| Radeon RX 5700 XT | AMD | RDNA | 8 GB GDDR6 | $262 | PLE Computers | Buy ↗ |
| GeForce GTX 1080 | NVIDIA | Pascal | 8 GB GDDR5X | $262 | PLE Computers | Buy ↗ |
| GeForce GTX 1660 SUPER | NVIDIA | Turing | 6 GB GDDR6 | $262 | PLE Computers | Buy ↗ |
| Arc A580 | Intel | Alchemist | 8 GB GDDR6 | $246 | PLE Computers | Buy ↗ |
| GeForce GTX 1070 | NVIDIA | Pascal | 8 GB GDDR5 | $185 | PLE Computers | Buy ↗ |
| Arc A380 | Intel | Alchemist | 6 GB GDDR6 | $169 | PLE Computers | Buy ↗ |
| Radeon RX 580 8GB | AMD | GCN 4 (Polaris) | 8 GB GDDR5 | $153 | PLE Computers | Buy ↗ |
| GeForce GTX 1060 6GB | NVIDIA | Pascal | 6 GB GDDR5 | $137 | PLE Computers | Buy ↗ |
Best workstation GPU prices
| GPU | Manufacturer | Architecture | VRAM | Best | Vendor | Buy |
|---|---|---|---|---|---|---|
| RTX 6000 Ada | NVIDIA | Ada Lovelace | 48 GB GDDR6 | $10,529 | PLE Computers | Buy ↗ |
| Radeon PRO W7900 | AMD | RDNA 3 | 48 GB GDDR6 | $5,418 | PLE Computers | Buy ↗ |
| Arc Pro B60 | Intel | Battlemage | 24 GB GDDR6 | $973 | PLE Computers | Buy ↗ |
Best data-centre GPU prices
| GPU | Manufacturer | Architecture | VRAM | Best | Vendor | Buy |
|---|
Explore
GPU prices
Every current GPU compared: manufacturer, architecture, VRAM and the best current price across shops, with buy links.
- NVIDIA: GeForce RTX, RTX workstation, data centre
- AMD: Radeon RX, Radeon PRO, Instinct
- Intel: Arc, Gaudi
- Laptops · Desktops
GPU hosting
Rent a GPU by the hour. Lowest advertised on-demand rates across the major clouds.
- H100 80 GB: from $1.65/hr (Vast.ai)
- A100 80 GB: from $0.80/hr (Vast.ai)
- RTX 4090: from $0.35/hr (Vast.ai)
- H200 141 GB: from $3.49/hr (Lambda)
- B200 192 GB: from $4.99/hr (Lambda)
AI
Providers vs open-source tools, by modality: everything that runs on a GPU.
- Text & LLMs: GPT, Claude, Llama, DeepSeek
- Image: Midjourney, FLUX, Stable Diffusion
- Video: Sora, Veo, HunyuanVideo, Wan
- Audio: Suno, ElevenLabs, MusicGen, Whisper
- Run it locally with Ollama, ComfyUI & vLLM
GPU programming
How to make the GPU do work, each with install steps and a runnable example.
- CUDA: NVIDIA, most mature
- ROCm / HIP: AMD, CUDA-portable
- Triton: Python GPU kernels
- WebGPU: any GPU, in the browser
- OpenCL: cross-vendor standard
- Vulkan · OpenGL · DirectX · WebGL
Latest news
Hardware2026-06-18
Blackwell B200 and GB200 NVL72 ramp across clouds
NVIDIA's Blackwell data-centre parts are now bookable by the hour, with 192 GB of HBM3e per GPU pushing single-node model sizes higher.
Source: nvidia.com
Image2026-06-12
Nano Banana Pro brings Gemini-3 image generation
Google DeepMind's Nano Banana Pro sets a new bar for in-image text and precise editing, intensifying the race with GPT Image and FLUX.
Source: blog.google
Image2026-06-05
Z-Image: efficient single-stream diffusion goes open
Tongyi-MAI's Z-Image shows high-quality generation from a compact single-stream architecture that runs comfortably on consumer GPUs.
Source: github.com
What is gpu.org?
A single, practical reference for everything GPU: what to buy and for how much, where to rent compute by the hour, how to program the hardware, and which AI models (hosted or open) run on it. Prices and listings are indicative, so always check the retailer for the latest price.