Mac mini dock or stand with an NVMe slot
Adds storage and front-facing ports without buying a larger internal drive. Verify the Mac model, Thunderbolt version, and power-delivery limits before ordering.
Search current listings on AmazonLocal model compatibility
Know before you download.
Can your Mac or PC run Qwen3.8, Gemma 4 or gpt-oss? Check whether a local AI model fits your machine, what to install, and where the estimate comes from.
Local device check
Let this browser suggest your machine.
Only your browser is checked. Nothing is uploaded.
Reading the local graphics adapter…
Comfortable local models at an 8K context
No catalog model currently clears the comfort threshold on this device.
The browser could not identify a reliable GPU. Choose the exact profile manually.
Compatibility check
Can your machine run it?
Pick a machine, a model, a quantization, and a context length. The result uses published weight sizes plus a conservative KV-cache estimate and 1.5 GB runtime overhead. It is not a benchmark.
Result
- Accelerator memory
- System memory
- Estimated requirement
- Weight files
- KV cache
- Accelerator headroom
- Fit ratio
Recommended hardware
The minimum comfortable upgrade target
As an Amazon Associate I earn from qualifying purchases. Product links open Amazon search results, may be affiliate links, and do not change the price you pay. Read the affiliate disclosure.
Low-cost setup upgrades
Accessories that solve the bottleneck before a GPU upgrade
These are generic Amazon search categories, not live product or price claims. Verify compatibility with the exact machine before buying.
External NVMe SSD for model storage
Keeps large GGUF and MLX files off the boot drive. Fast storage reduces load time, but it does not increase the memory available for inference.
Search current listings on AmazonThunderbolt or USB4 NVMe enclosure
A reusable enclosure lets you choose the drive capacity separately. Confirm the host supports the advertised Thunderbolt or USB4 generation.
Search current listings on AmazonAs an Amazon Associate I earn from qualifying purchases. Links are sponsored search links and do not change the price you pay. Affiliate disclosure.
Low-cost setup upgrades
Accessories that solve the bottleneck before a GPU upgrade
These are generic Amazon search categories, not live product or price claims. Verify compatibility with the exact machine before buying.
External NVMe SSD for model storage
Keeps large GGUF and MLX files off the boot drive. Fast storage reduces load time, but it does not increase the memory available for inference.
Search current listings on AmazonDesktop RAM upgrade kit
Useful for MoE CPU offload and keeping the operating system out of VRAM. Match the motherboard's memory type, slot layout, capacity limit, and validated part numbers.
Search current listings on AmazonGPU support bracket
A lower-cost mechanical upgrade for heavy three-slot cards. Check clearance, mounting height, and whether the bracket blocks fans or case panels.
Search current listings on AmazonAs an Amazon Associate I earn from qualifying purchases. Links are sponsored search links and do not change the price you pay. Affiliate disclosure.
Shop by memory
Start with the VRAM tier you actually have.
Popular pairings
One model, one machine, one answer.
Flint 40B on NVIDIA GeForce RTX 5090 · 32 GB
20.2 GB estimated · UD-Q3_K_XL
NVIDIA Nemotron 3.5 Bolt E4B on Mac mini M6 · 24 GB
8.0 GB estimated · Q4_K_M
LFM3-1.7B on Mac mini M4 · 16 GB
3.4 GB estimated · Q8_0
Bonsai 2 on Mac mini M6 · 24 GB
9.2 GB estimated · PQ2_0
Qwen3.8 27B on NVIDIA GeForce RTX 5060 Ti · 16 GB
19.4 GB estimated · Q4_K_M
Gemma 4 12B Unified on Mac mini M6 · 24 GB
14.7 GB estimated · Q8_0
Gemma 4 E4B on AMD Radeon RX 9070 XT · 16 GB
11.0 GB estimated · Q8_0
gpt-oss-20b on Mac Studio M5 Ultra · 96 GB
13.3 GB estimated · Q4_K_M
Qwen3.6 35B-A3B on NVIDIA GeForce RTX 5090 · 32 GB
24.7 GB estimated · Q4_K_M
Gemma 4 31B on NVIDIA DGX Spark · 128 GB
36.5 GB estimated · Q8_0
GLM-5.3-Flash on Mac Studio M5 Ultra · 256 GB
203.2 GB estimated · UD-Q4_K_XL
GLM-4.7-Flash on NVIDIA GeForce RTX 5090 · 32 GB
23.4 GB estimated · Q5_K_M
Apple silicon
Unified memory, without the hand-waving.
Mac mini M6 · 16 GB
16 GB unified memory · 153 GB/s
Mac mini M5 Pro · 24 GB
24 GB unified memory · 307 GB/s
Mac mini M6 · 24 GB
24 GB unified memory · 170 GB/s
Mac mini M6 · 32 GB
32 GB unified memory · 170 GB/s
New models
Fresh local releases, checked before they trend.
GLM-5.3-Flash
Z.ai · MIT
GLM-4.7-Flash
Z.ai · MIT
Flint 40B
Recursive · Apache 2.0
NVIDIA Nemotron 3.5 Bolt E4B
NVIDIA · OpenMDW-1.1
LFM3-1.7B
Liquid AI · LFM Open License v1.0
Bonsai 2
Prism ML · Apache 2.0
GLM-4.5-Air
Z.ai · MIT
Qwen3.8 27B
Alibaba Cloud · Apache 2.0
Gemma 4 12B Unified
Google DeepMind · Apache 2.0
gpt-oss-20b
OpenAI · Apache 2.0
GLM-5.3
Z.ai · GLM-5.3 License
DeepSeek V4.1 Flash
DeepSeek · MIT
Curious whether the 552B DeepSeek V4.1 Flash can run locally? Read the server-tier reality check.
Hardware
Compare the machine you have.
Mac mini M6 · 24 GB
24 GB unified memory
NVIDIA GeForce RTX 5060 Ti · 16 GB
16 GB VRAM
AMD Radeon RX 9070 XT · 16 GB
16 GB VRAM
NVIDIA DGX Spark · 128 GB
128 GB shared CPU/GPU memory
Method
Why the numbers can be trusted.
Every model page lists its source and last-verified date. Minimum memory is a conservative starting point, not a guarantee. If a number is unknown, the page says so instead of inventing precision.
The checker does not send your selections anywhere. It runs in your browser. Read the formula, thresholds, sources, and limitations.