Local AI Stats.
Skip to content

COMPUTER RECORD · SOURCED SPECIFICATIONS

AMD Ryzen AI Halo: specifications and local LLMs

128 GB unified memory and 256 GB/s rated memory bandwidth. The AMD Max+ 395 model specifies 128GB unified memory and an x86 AI software platform.

Catalog reviewed . Listing evidence: 2026-09-24. Price and performance records retain their own dates.

As an Amazon Associate I earn from qualifying purchases.

CPU, GPU, memory and storage

CPU
Ryzen AI Max+ 395 · 16 cores / 32 threads
GPU
Radeon 8060S · 40 CUs
Installed memory
128 GB · LPDDR5X unified · Manufacturer model; seller configuration unconfirmed
Unified memory, not separate VRAM
128 GB shared with CPU and operating system
Memory bandwidth
256 GB/s rated / theoretical
Storage
2000 GB · Manufacturer model; seller configuration unconfirmed
Operating system
Linux (screenshot)
Expansion
Manufacturer model: M.2 / 10GbE; fixed memory
NPU advertised peak
50 TOPS · not LLM tokens/s
Whole-chip advertised peak
126 TOPS · distinct from NPU-only performance

The screenshot does not show the exact seller CPU, RAM or SSD. Confirm these before comparing value or buying.

The screenshot confirms the Halo developer desktop, Linux and displayed price. The Max+ 395 / 128GB / 2TB specifications are AMD model specifications; the seller configuration is still unconfirmed.

Owner-supplied Amazon.com screenshot reviewed 24 Sep 2026 and matched by the owner to ASIN B0H5RNZ43D. The screenshot shows the product title and price, but not the ASIN or capture time. It is not an independent live listing check. Exact seller CPU, RAM and SSD remain unconfirmed. Current seller, stock and checkout price can change.

What local models could AMD Ryzen AI Halo run?

The 128 GB unified-memory pool gives a 89.6 GB initial weight budget using our 30% (minimum 2 GB) reserve. The CPU, GPU and OS share this capacity; software allocation limits still apply.

129 catalog files pass this size check. The largest is Xor 1.1 at 70.214 GB (BF16). This is a catalog weight limit, not a measured maximum practical model size. Context length, runtime and architecture support can rule out a candidate.

Model snapshot: 14 October 2026. The examples below use MMLU-Pro; models without that score are not ranked in this table.

MMLU-Pro memory candidates · GB = 1 billion bytes. Published scores may use different settings or precision.
Model / quantizationWeightsParametersScore / 100Score / GBBudget leftSources
Qwen3.5-4BQ4_K_M · Pareto choice2.741 GB4.206B79.128.85986.859 GBGet weights
Score source
MiniCPM5 2BQ4_K_M · Pareto choice1.561 GB2.517B70.845.34688.039 GBGet weights
Score source
Qwen3.5-9BQ4_K_M · Pareto choice5.681 GB9.653B82.514.52383.919 GBGet weights
Score source
LFM2.5 230MQ4_K_M · Pareto choice0.153 GB0.23B20.25132.00289.447 GBGet weights
Score source
Qwen3.5-27BQ4_K_M · Pareto choice16.741 GB27.781B86.15.14372.859 GBGet weights
Score source

Choose this computer in the memory tool · Compare its memory tier

Published LLM performance

Local AI Stats has not tested this computer. These third-party results apply to their stated workloads.

GPT-OSS 20B · vLLM · batch 64

617 aggregate tokens/s

StorageReview’s firsthand Linux vLLM serving test: 256 input / 256 output tokens, batch 64. Aggregate throughput across requests, not single-user chat speed. Quantization and exact software versions are not disclosed in the text; treat this as a reported comparison of two tested systems.

Original test source

GPT-OSS 120B · vLLM · batch 64

222 aggregate tokens/s

Same StorageReview serving suite: Linux vLLM, 256 input / 256 output tokens, batch 64. Aggregate throughput; quantization and exact software versions are not disclosed in the text. Results apply only to the two tested systems.

Original test source

Advertised compute peaks and bandwidth are not tokens per second. Compare the same model, quantization, context and concurrency.

Price and value

ASIN: B0H5RNZ43D · Amazon.com listing

$4,299.99 USD reference. Amazon.com price shown in an owner-supplied screenshot reviewed 24 Sep 2026; capture time not provided. Excludes shipping, import charges, tax and optional discounts. This is a dated reference, not a live checkout quote. RAM, SSD and exact processor are not visible; the displayed model capacities come from AMD, not this offer.

Reference price per installed-memory GB: $33.594. This measures cost per GB, not speed or total running cost.

As an Amazon Associate I earn from qualifying purchases. Check the exact configuration, seller, marketplace and current price before buying.

GET NOW · Amazon

Related hardware

Compare all hardware · Compare budgets

Specification and price sources