Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsChoose the computer around the model and workflow you will actually run. Start by identifying the model size, runtime and operating system you need; then check usable GPU or unified memory, exact GPU power and cooling, storage, upgrade options and whether you need to carry the machine. A laptop is the right answer when mobility and a built-in screen matter. A compact AI desktop fits a permanent workspace when its memory architecture and software stack support your models. A configurable desktop is usually the most flexible choice when you need a particular GPU, expansion room or future upgrades.
NVIDIA Developer summarizes the decision this way: “Choose hardware based on operating system, available GPU or unified memory, model size, and workflow.” That is a better starting point than labels such as “AI PC,” NPU TOPS or a processor family name.
As an Amazon Associate I earn from qualifying purchases.
Start with the workload, not the form factor
“Local AI” can mean very different things: running a small assistant without sending prompts to a server, experimenting with image-generation models, embedding documents for search, fine-tuning a model, serving an internal chatbot, or developing and testing software. Each task stresses hardware differently. A machine that is excellent for one may be a poor fit for another.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Write down the models and runtimes
- List the model families, parameter sizes and quantization formats you intend to use. Include the largest model you expect to run, not only the first one you will download.
- Record the runtime and framework: for example, a vendor GPU stack, a CPU-only runtime, a platform-specific accelerator framework or a containerized Linux environment.
- Note whether you need training or fine-tuning, several models loaded at once, long context windows, image or video generation, or only short interactive prompts.
- Decide whether the computer must work offline. Offline use increases the importance of local storage, model-management tools and update procedures.
Do not buy on the assumption that a future software update will make an incompatible model run well. Check the model’s supported operating systems, accelerator back ends, instruction sets and memory requirements before comparing computers.
#1 Best Overall
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
Separate capacity from speed
Memory capacity determines whether a model and its working data can fit. Compute throughput, memory bandwidth, thermal limits and software kernels determine how quickly it runs once it fits. A system with more nominal memory is not automatically faster, and a high peak-compute number does not tell you the tokens per second for your particular model.
The five checks that decide the purchase
- Model fit: estimate the model weights, runtime overhead, context or image buffers and operating-system headroom together.
- Memory architecture: distinguish dedicated GPU memory from unified system memory and check how the selected software uses it.
- Software support: verify the operating system, drivers, framework, accelerator back end and container support for every important workload.
- Sustained operation: check the exact power limit, cooling design, fan behavior and ability to run for the duration of your jobs.
- Lifecycle: compare portability, upgrade paths, storage replacement, repairability, warranty and the cost of replacing the whole system if one component becomes limiting.
These checks matter more than a badge that says “AI.” An NPU, for example, may accelerate selected operating-system features while your local language model still depends on GPU memory, unified memory or CPU resources.
How the three form factors compare
| Decision axis | Laptop | AI mini PC or compact AI desktop | Configurable desktop |
|---|---|---|---|
| Portability | Strongest: screen, keyboard and battery are integrated. | Low; normally remains on a desk and needs an external display and peripherals. | Low; requires a permanent setup and separate peripherals. |
| Memory and accelerator choice | Depends on the exact GPU, system memory and power configuration. | Some specialized systems offer large unified-memory pools; architecture and supported software must be checked. | You select the GPU, memory, motherboard, storage and power supply to match the workload. |
| Sustained performance | Constrained by the model’s power and cooling design; the same GPU family can perform differently in different laptops. | Depends on the compact chassis, cooling and workload; vendor claims are not independent comparisons. | More room for cooling and power, but performance still depends on the installed components and build quality. |
| Upgradeability | Often limited. Verify whether memory, storage and the GPU are replaceable before buying. | Varies widely; many compact systems have soldered memory or proprietary modules. | Usually the broadest choice of GPU, storage, memory and power-supply upgrades, subject to case and motherboard limits. |
| Best question | Will I carry it, and does this exact GPU have enough usable memory? | Does this system’s memory architecture and software stack support my target models? | Do I need a specific GPU or a path to upgrade individual components? |
This is a decision framework, not a performance ranking. The available documentation does not provide a controlled, independent benchmark running identical models and software across all three categories.
Free tools Windows power users keep installed
One-click scans. No signup required.
Choosing an AI mini PC or compact AI desktop
When a compact system makes sense
Choose a mini PC when the machine will stay at one desk, you value a small footprint and you have confirmed the architecture used by your models. It can be quieter and simpler than assembling a tower, and it avoids paying for a laptop screen, keyboard and battery that will never leave your office. You still need to budget for a monitor, keyboard, mouse, networking and possibly external storage.
Unified memory needs careful interpretation
Some compact AI systems use coherent unified memory shared by the CPU and accelerator. NVIDIA’s DGX Spark product information lists 64 GB and 128 GB configurations and describes the platform as a compact local development and inference system. Those are vendor-published capacity figures, not an independent speed test. Unified memory is not automatically equivalent to the same amount of dedicated GPU VRAM: bandwidth, software allocation, reserved system memory and the accelerator architecture all affect usable capacity and speed.
Apple’s Mac mini (2024) illustrates a different unified-memory desktop stack. Apple lists 16 GB as a base configuration for the M4 version, configurable to 24 GB or 32 GB, while the M4 Pro version starts at 24 GB and can be configured to 48 GB or 64 GB. These figures describe that 2024 generation and should not be treated as specifications for a newer model or as a direct performance comparison with NVIDIA hardware.
Questions to answer before buying
- Which operating systems and runtimes are officially supported on the architecture?
- Can the accelerator run your preferred quantization, attention implementation and image or video kernels?
- How much memory remains after the operating system and desktop services reserve their share?
- Can you replace or expand storage, and is there a practical way to attach fast external storage?
- What happens when a long job runs for hours: does the chassis sustain its advertised performance, reduce clocks or become too loud for your workspace?
NVIDIA states that DGX Spark can deliver up to 1 PFLOP of AI performance at FP4 precision. That is a vendor-stated peak for a particular precision and workload class; it is not a promise of a specific model’s tokens per second, nor an apples-to-apples comparison with a laptop or tower.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchChoosing a laptop for local AI
Why a laptop is attractive
A laptop is the practical choice when you need to move between home, work, class or travel and want one integrated computer. It includes the display, keyboard, battery, webcam and often a power-management system that can switch between quiet battery operation and plugged-in performance.
Check the exact GPU implementation
Laptop GPU names hide important differences. NVIDIA’s laptop comparisons show that memory configurations and power ranges vary among models in the same GPU family. Two laptops carrying the same family name can therefore have different performance, fan noise and battery behavior. Read the manufacturer’s specification for the exact SKU: GPU memory capacity and type, graphics power range, cooling design, charger rating and whether the GPU is enabled when the system is on battery.
Rank #2
- Built for Local AI Development: AMD Ryzen AI Halo is designed for local AI development and inference, featuring 128GB unified memory and support for up to 200B parameter models to build and run intensive AI workloads locally.
- 128GB Unified Memory: Features 128GB LPDDR5x unified memory at 8000 MT/s with 256 GB/s memory bandwidth, providing a shared memory pool across the CPU, GPU, and NPU to support larger AI models.
- AMD Ryzen AI Max+ 395 Processor: Features 16 cores, 32 threads, and Zen 5 architecture, paired with AMD Radeon 8060S integrated graphics featuring 40 RDNA 3.5 compute units and an AMD XDNA 2 NPU with up to 50 TOPS.
- Linux AI Developer Platform: Purpose-built for Linux-based AI development with full AMD ROCm software support and preloaded tools, models, and workflows optimized for local AI development.
- Compact, Connected Design: Includes a 2TB M.2 SSD, 10GbE LAN, Wi-Fi 7, Bluetooth 5.4, USB-C connectivity, and HDMI 2.1b.
Also verify whether system memory is replaceable. A laptop with adequate memory today can become a dead end if the model, operating system and other applications consume most of it and the modules are soldered.
Plan for plugged-in workloads
Large local models, image generation and fine-tuning can drain a battery quickly and may run substantially slower on battery power. Test or confirm the vendor’s performance mode, charger requirement, fan profile and external-display behavior. If you routinely run overnight jobs, a desk-oriented system may provide better acoustics, cooling and component longevity than a thin laptop.
Choosing a configurable desktop
What a tower buys you
A conventional desktop lets you choose the exact GPU, dedicated GPU memory, CPU, system memory, storage and power supply. NVIDIA’s local-AI guidance discusses GeForce RTX and RTX PRO options for desktop and laptop development. For a desktop, that flexibility is the main advantage: you can prioritize a larger GPU memory pool, add fast storage, install more system memory or replace one limiting component instead of replacing the entire computer.
Match the supporting components
- Power supply: size it for the selected GPU and future upgrades, with the required connectors and adequate sustained output.
- Case and cooling: provide unobstructed intake and exhaust, a cooler rated for the CPU and enough room for the GPU’s length and thickness.
- Motherboard: check PCIe slots, memory capacity, M.2 sockets, networking and firmware support.
- Operating system: choose the OS that has first-class drivers and framework support for your target accelerator.
- Storage: leave room for model files, caches, checkpoints, datasets and temporary outputs; a nearly full drive can make updates and swap operations unreliable.
A tower is not automatically faster than every laptop or compact system. A poorly selected GPU, insufficient power, thermal throttling or an unsupported software stack can erase the theoretical advantage.
GPU memory, unified memory and model capacity
Use a memory budget, not a single headline number
Build a budget that includes model weights, runtime allocations, tokenizer and context buffers, operating-system memory, graphics or display use, and any second model or dataset you keep resident. Leave practical headroom for longer prompts and software updates. If the model barely fits, startup may succeed but generation can become unstable when the context grows.
Dedicated GPU memory
Dedicated VRAM is physically attached to the GPU and is normally the fastest place for accelerator workloads. Capacity is a hard constraint for many runtimes. When a model spills into system memory, performance can change sharply because data must cross a slower link and compete with the CPU and operating system.
Recommended Free Tools
Unified memory
Unified memory allows CPU and accelerator components to address a shared pool. That can simplify data movement and make a large capacity useful for models that would not fit in a smaller discrete GPU. It does not guarantee the same bandwidth, kernel support or speed as a discrete GPU with an equal capacity. Compare the architecture and supported software, not just the number of gigabytes printed on the box.
Do not confuse memory with compute
A machine can have enough memory to load a model yet generate slowly because its accelerator, bandwidth or cooling is inadequate. Conversely, a very fast GPU with insufficient memory cannot run the model at the desired settings. Capacity answers “can it fit?”; compute and software answer “how fast and reliably will it run?”
NPU TOPS: useful for features, not a universal local-LLM score
Microsoft’s Copilot+ PC developer guidance says many Copilot+ features require an NPU capable of more than 40 trillion operations per second (TOPS). That threshold describes eligibility for a class of Windows on-device features. It is not a general benchmark for language-model size, context length or tokens per second.
Rank #3
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 64GB pool, which is perfect for running LLMs such as Deepseek 32B, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 4% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
When evaluating an NPU laptop, ask three separate questions:
- Does the operating system feature I want explicitly require an NPU, and does this device meet the current requirement?
- Does my model runtime support this NPU, including the operators and quantization format I use?
- If the NPU is unsupported or too limited, does the laptop still have enough CPU, GPU or unified memory to run the workload acceptably?
A Copilot+ classification therefore does not mean that every local model will run on the NPU, or that the computer has enough memory for a large model.
Operating-system and framework compatibility
Software fit can outweigh a small hardware difference. Before purchase, confirm:
- Driver availability and update policy for the GPU, NPU or integrated accelerator.
- Support for your operating system edition, kernel or subsystem and container tools.
- Whether your runtime uses the accelerator by default or needs a specific build, environment variable or backend.
- Support for the model’s attention kernels, quantization, image codecs and other operators.
- Whether you can install development tools, compilers and libraries without workarounds that complicate updates.
If a project’s documentation lists only one platform, treat ports and compatibility layers as experiments rather than purchase guarantees. A mainstream desktop GPU may offer broader community support, while a specialized compact platform may provide a tightly integrated stack but fewer third-party examples.
Cooling, noise and sustained performance
Local AI is often a sustained load, unlike a short benchmark burst. A laptop may reduce power after its heat pipes saturate. A mini PC may protect a small chassis by lowering clock speeds. A tower can sustain higher power only if its cooler, case airflow and power supply are designed for it.
Ask for sustained, not peak, behavior
- Look for the GPU’s configurable power range and the CPU’s sustained thermal rating.
- Check whether performance modes require the system to be plugged in.
- Consider fan noise in the room where you will work or sleep.
- Make sure the system can run your longest job without filling its drive or exhausting memory.
Vendor peak figures, including the DGX Spark claim of up to 1 PFLOP at FP4, should be read in their stated precision and workload context. They are not substitutes for a benchmark using your model and settings.
Storage, networking and desk requirements
Model files, multiple quantizations, checkpoints and datasets can consume storage faster than expected. Prefer a fast internal drive with free space for temporary files and updates. Confirm whether the mini PC or laptop allows a second drive; on a desktop, reserve an M.2 slot or drive bay for growth.
Networking matters when you download large models, access a remote dataset or serve the model to other devices. Wired Ethernet can provide a steadier connection for a permanent workstation, while a laptop benefits from reliable Wi-Fi and a practical way to connect external displays and storage. A mini PC usually needs a monitor, keyboard, mouse, webcam and speakers in addition to the computer itself; include those in the total cost and desk space.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Total cost and upgrade strategy
Compare the complete system, not only the computer’s advertised price. Include memory and storage upgrades, a suitable charger or power supply, display and peripherals, operating-system licensing where applicable, electricity for sustained workloads and the cost of replacing a non-upgradable machine.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Rank #4
- 【Leading AI Mini Workstation】MINISFORUM AI MS-S1 Max Workstation comes with AMD Ryzen AI Max+ 395 processor, which uses AMD's latest generation Zen 5 architecture. It has 16 Cores and 32 Threads, the boost clock is up to 5.1GHz. The overall processor performance is up to 126 TOPS, and the NPU performance reaches up to 50 TOPS. AMD Ryzen AI enables improved productivity, advanced collaboration, and improved efficiency.
- 【AMD Radeon 8060S Graphics 】The MS-S1 Max Mini PC equipped with AMD Radeon 8060S Graphics which built on the new generation of RDNA 3.5 architecture AMD graphics, it brings ultra-high frame rate experiences and advanced content creation features anywhere and delivers staggering performance. It can handle all your computing and multimedia tasks efficiently.
- 【Five 8K Video Output】This MS-S1 Max Workstation comes with five video outputs, 1x HDMI (8K@60Hz), 2x USB4(40Gbps,Alt DP2.0,PD out 15W) and 2x USB4 V2(80Gbps,Alt DP2.0,PD out 15W) Outputs, which support multiple monitors display at the same time and provide a larger and wider filed of view and improve your work efficiency. It is used in fields that require high-performance computing and graphics processing, including digital signage and securities trading, as well as work that uses CAD, such as engineering design, scientific calculations, animation production, and post-production for movies and television
- 【 Fast and Stable Wire & Wireless Speed】It comes with Two 10G Lan Ports for wired connection and and Wi-Fi 7 / BT5.4 for wireless connection, which increased the network speed greatly and expand its functions and improved performance of computer to a large extent and allows you to use more networks such as software routers (OpenWRT / DD-WRT / Tomato etc.), firewalls, NAT, network isolation etc.
- 【Large Storage & Flexible Expandability】This Workstation equipped with 64GB LPDDR5-8000MHz + 2TB M.2 2280 PCIe4.0 SSD. There is another PCIe4.0 SSD slot available for up to 8TB, these SSD slots are compatible with RAID0 and RAID1, you can store movies, videos, photos, important files easily. What’s more, it also comes with 1x standard PCIex16 slot(PCIe4.0x4) inside.
Choose a replacement cycle deliberately
- Short, mobile cycle: a laptop can justify a higher initial cost if one device replaces a computer and tablet for your work.
- Component cycle: a desktop can be economical when you expect to upgrade the GPU or storage separately.
- Specialized cycle: a compact AI platform may be worthwhile when its integrated memory and software stack solve a specific workload that would require several add-on components elsewhere.
Retail prices, availability, exact configurations and program eligibility change over time. Verify the current SKU and warranty at purchase rather than relying on a generic product-family page.
Decision paths for common buyers
Choose a laptop when
- You regularly carry the computer and need a built-in display, keyboard and battery.
- The exact laptop GPU has sufficient memory and an appropriate power range for your largest model.
- You accept less upgradeability and potentially lower sustained performance than a well-cooled tower.
Choose a compact AI desktop when
- The computer will stay at one desk and you want a small, quiet setup.
- Its unified-memory capacity and architecture fit your target models.
- Your frameworks and runtimes explicitly support the platform.
Choose a configurable desktop when
- You need a particular GPU memory capacity, expansion card or storage layout.
- You expect to upgrade components or run long, sustained jobs.
- You can provide the desk space, peripherals, cooling and power that a tower requires.
Choose an NPU-focused laptop when
- Your priority is a defined set of supported Windows on-device features.
- You have verified the current NPU requirement and feature support.
- You have separately confirmed how your local models will use CPU, GPU or unified memory.
A purchase checklist
- Write the largest model, quantization, context length and runtime you plan to use.
- Calculate a memory budget that includes the model, runtime, operating system and headroom.
- Identify whether the workload requires dedicated VRAM, unified memory, CPU RAM or a supported NPU backend.
- Confirm the exact GPU or accelerator model, memory capacity and laptop power configuration.
- Verify operating-system, driver, framework, container and model-operator support.
- Check sustained cooling, fan noise, charger or power-supply requirements and external-display behavior.
- Price storage, peripherals, upgrades, electricity and warranty support as part of the complete system.
- Decide which component you may need to replace first and whether the chosen form factor permits that replacement.
- Before the return period ends, install the intended runtime, load a representative model and monitor memory use, speed, temperature and stability.
How to validate the computer after purchase
Use your real workflow rather than a generic score. Install the production runtime, load the largest model you expect to use, exercise the longest context or image size, and run a sustained session while watching memory allocation, clock behavior, temperatures and fan noise. Repeat on battery and on AC power if you bought a laptop. Confirm that the accelerator is actually being used; a successful launch on the CPU alone can look like compatibility while delivering unusable speed.
Keep the test results with the system’s exact BIOS, driver and runtime versions. When an update changes performance, you can identify whether the change came from software, power settings or the model configuration.
Common mistakes and recovery steps
Buying by processor or GPU family name
Symptom: the computer is much slower than another system with the same family label. Fix: compare the exact laptop GPU power range, memory configuration, cooling and driver version.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Treating unified memory as identical to VRAM
Symptom: a model fits but runs slowly or competes with normal desktop applications. Fix: check usable memory, bandwidth and runtime support; close competing applications and use a model or quantization that fits with headroom.
Using NPU TOPS as a model-capacity promise
Symptom: a Copilot+ laptop supports system features but your preferred local runtime ignores the NPU. Fix: verify backend support and evaluate the GPU, CPU and memory independently.
Ignoring sustained thermals
Symptom: the first minutes are fast, followed by lower speed or loud fans. Fix: use the manufacturer’s performance mode and charger, improve airflow, limit the workload or choose a system with more cooling capacity.
Filling the system drive
Symptom: model downloads fail, updates break or generation becomes unstable. Fix: move models and caches to a fast drive with free space and retain an explicit reserve for the operating system.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
