LLMBOTTLENECK.COM
AMD / unified

AMD Ryzen AI Max+ 395 with Radeon 8060S

128 GB decides what fits. 256 GB/s decides how fast it runs once it does.

Computed for 128 GB, the largest configuration. The AMD Ryzen AI Max+ 395 with Radeon 8060S is also sold with 64 GB, and what fits changes with it.

Run it with 64 GB →

Open in the calculator →Best models for 128 GB, on every card that size →

Find one: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Quick answer
The AMD Ryzen AI Max+ 395 with Radeon 8060S fits 248 of 320 sized models entirely in its 128 GB; the largest widely used one is Qwen3.8-Flash-Next (111.6 GB at Q4_K_M). Memory decides what fits; its 256 GB/s of bandwidth decides how fast it answers.

At 8,192 tokens of context with the whole model in device memory. Speeds are estimates from memory bandwidth, not benchmarks run on this card.

AMD Ryzen AI Max+ 395 with Radeon 8060S · 128 GB · 256 GB/s · Q4_K_M where published · 8,192 tokensSpecificationpublished dataDecode speedestimate

248 of 320 fit entirely

248 of the 320 models the engine can size fit entirely in device memory at Q4_K_M where it is published, otherwise the nearest published format, and 8,192 tokens. Unified memory is one pool, so there is no second memory tier to spill into; the remaining 72 do not run at this context.

78%of the models the engine can size fit entirely
128GB of device memory
256GB/s memory bandwidth
0run with system memory
72do not run at 8,192 tokens

Featured models · Q4_K_M at 8,192 tokens, including offload

ModelNeedsVerdictDecodeCalculator
Qwen3.5-2B2.3B parameters2.32 GBfits~162 tok/sOpen →
NVIDIA-Nemotron-3-Nano-4B-BF164.0B parameters3.46 GBfits~75 tok/sOpen →
Qwen3-VL-8B-Instruct8.8B parameters7.04 GBfits~34 tok/sOpen →
Qwen3.5-9B9.7B parameters7.05 GBfits~41 tok/sOpen →
gemma-4-12B-it12B parameters9.54 GBfits~24 tok/sAbout reading paceOpen →
gpt-oss-20b21B parameters13.8 GBfits~80 tok/sOpen →
Qwen3.8-27B28B parameters18.6 GBfits~13 tok/sAbout reading paceOpen →
Qwen3.6-27B28B parameters18.6 GBfits~13 tok/sAbout reading paceOpen →
Kimi-Linear-48B-A3B-Instruct49B parameters30.4 GBfits~107 tok/sOpen →
Qwen3.8-Flash-Next180B parameters111.6 GBfits~55 tok/sOpen →
DeepSeek-V4-Flash-Vision-Exp305B parameters188.1 GBshort by 60.1 GBdoes not runOpen →
GLM-5.3-Flash321B parameters198.3 GBshort by 70.3 GBdoes not runOpen →

These examples are selected from prominent labs using the catalogue’s latest Hugging Face 30-day downloads and repository-creation freshness signal, with newer releases guaranteed a place. Offloaded rows assume 32 GB of system RAM, and a speed is only shown for a row that runs. Fitting in memory is not the same as loading: whether the runtime and version you have supports each architecture and format on this machine has not been tested here. Each “Open” link carries the same model, format, context and RAM into the calculator.

Fits entirely in AMD Ryzen AI Max+ 395 with Radeon 8060S (128 GB) memory — 248 of 320 sized models

Every model that fitswhole model resident · offload off

The most demanding model that fits is Step-3.5-Flash at Q4_K_M: 127.0 GB of the 128 GB, leaving 1.03 GB spare.

Fully resident at 8,192 tokens (or the model’s own maximum, where that is shorter), offload off, at 256 GB/s. Each row is a run of the engine for this configuration; the rows start with current, prominent releases and “fits” is memory, not a tested runtime. Older or less prominent models remain available through this search and “Show all”.

Showing 40 of 248 models that fit.

ModelFormatNeedsSpareDecodeDownloads / 30dCalculator
Qwen3.8-27BQwen · 28B paramsQ4_K_M18.6 GB109.4 GB~13 tok/sAbout reading pace6.9MOpen →
Qwen3.8-Flash-NextNEWQwen · 180B paramsQ4_K_M111.6 GB16.4 GB~55 tok/s1.4MOpen →
gemma-4-26B-A4B-itGoogle · 26B paramsQ4_K_M16.8 GB111.2 GB~75 tok/s13MOpen →
gemma-4-31B-itGoogle · 31B paramsQ4_K_M22.2 GB105.8 GB~9.7 tok/sSlow9.9MOpen →
Qwen3.5-9BQwen · 9.7B paramsQ4_K_M7.05 GB121.0 GB~41 tok/s9MOpen →
Qwen3.5-4BQwen · 4.7B paramsQ4_K_M3.98 GB124.0 GB~74 tok/s7.8MOpen →
Qwen3.6-35B-A3BQwen · 36B paramsQ4_K_M23.1 GB104.9 GB~108 tok/s3.3MOpen →
gemma-4-12B-itGoogle · 12B paramsQ4_K_M9.54 GB118.5 GB~24 tok/sAbout reading pace1.9MOpen →
Qwen3.6-27BQwen · 28B paramsQ4_K_M18.6 GB109.4 GB~13 tok/sAbout reading pace2.5MOpen →
Qwen3.5-2BQwen · 2.3B paramsQ4_K_M2.32 GB125.7 GB~162 tok/s4.9MOpen →
gemma-4-E4B-itGoogle · 8.0B paramsQ4_K_M5.86 GB122.1 GB~66 tok/s4.4MOpen →
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16NVIDIA · 32B paramsQ4_K_M20.3 GB107.7 GB~10 tok/sAbout reading pace530KOpen →
NVIDIA-Nemotron-3-Nano-4B-BF16NVIDIA · 4.0B paramsQ4_K_M3.46 GB124.5 GB~75 tok/s3.4MOpen →
gemma-4-E2B-itGoogle · 5.1B paramsQ4_K_M4.02 GB124.0 GB~174 tok/s3MOpen →
Qwen3.5-0.8BQwen · 873M paramsQ4_K_M1.46 GB126.5 GB~354 tok/s2.6MOpen →
Qwen3-VL-8B-InstructQwen · 8.8B paramsQ4_K_M7.04 GB121.0 GB~34 tok/s14.6MOpen →
Qwen3.5-27BQwen · 28B paramsQ4_K_M18.6 GB109.4 GB~13 tok/sAbout reading pace1.9MOpen →
North-Micro-Vision-InstructCohereLabs · 2.5B paramsQ4_K_M2.99 GB125.0 GB~106 tok/s180.7KOpen →
Qwen3.5-35B-A3BQwen · 36B paramsQ4_K_M23.1 GB104.9 GB~108 tok/s1.6MOpen →
NVIDIA-Nemotron-3-Super-120B-A12B-BF16NVIDIA · 124B paramsQ4_K_M76.9 GB51.1 GB~2.6 tok/sVery slow1.2MOpen →
granite-4.2-8bIBM · 8.8B paramsQ4_K_M7.49 GB120.5 GB~31 tok/s128.4KOpen →
GLM-4.7-Flashzai-org · 31B paramsQ4_K_M20.4 GB107.6 GB~10 tok/sAbout reading pace1.8MOpen →
granite-4.1-3bIBM · 3.4B paramsQ4_K_M3.72 GB124.3 GB~72 tok/s521.8KOpen →
LFM2.5-2.6BLiquidAI · 2.7B paramsQ4_K_M2.61 GB125.4 GB~115 tok/s108.1KOpen →
Qwen3-VL-4B-InstructQwen · 4.4B paramsQ4_K_M4.72 GB123.3 GB~54 tok/s3.5MOpen →
granite-4.1-30bIBM · 29B paramsQ4_K_M20.4 GB107.6 GB~10 tok/sAbout reading pace301.5KOpen →
Qwen3.5-122B-A10BQwen · 125B paramsQ4_K_M77.9 GB50.1 GB~39 tok/s512.6KOpen →
Qwen3-VL-2B-InstructQwen · 2.1B paramsQ4_K_M3.05 GB125.0 GB~98 tok/s2.8MOpen →
granite-4.2-3bIBM · 3.7B paramsQ4_K_M3.72 GB124.3 GB~72 tok/s50.6KOpen →
LFM2.5-230MLiquidAI · 230M paramsQ4_K_M1.05 GB126.9 GB~805 tok/s88.6KOpen →
Qwen3-0.6BQwen · 752M paramsQ4_K_M2.22 GB125.8 GB~150 tok/s29.7MOpen →
Qwen3-Coder-NextQwen · 80B paramsQ4_K_M50.0 GB78.0 GB~72 tok/s596.3KOpen →
gpt-oss-20bOpenAI · 21B paramsQ4_K_M13.8 GB114.2 GB~80 tok/s6.6MOpen →
granite-4.2-30bIBM · 29B paramsQ4_K_M20.7 GB107.3 GB~10 tok/sAbout reading pace35.8KOpen →
granite-4.1-8bIBM · 8.8B paramsQ4_K_M7.49 GB120.5 GB~31 tok/s179.3KOpen →
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16NVIDIA · 32B paramsQ4_K_M20.3 GB107.7 GB~10 tok/sAbout reading pace875.2KOpen →
gpt-oss-120bOpenAI · 117B paramsQ4_K_M71.9 GB56.1 GB~56 tok/s4.5MOpen →
LFM2.5-VL-3BLiquidAI · 3.1B paramsQ4_K_M2.85 GB125.1 GB~115 tok/s26.7KOpen →
MiMo-V2.6-Distill-Qwen-9BNEWXiaomiMiMo · 9.4B paramsQ4_K_M6.90 GB121.1 GB~41 tok/s14.6KOpen →
Qwen3-4B-Instruct-2507Qwen · 4.0B paramsQ4_K_M4.72 GB123.3 GB~54 tok/s3.7MOpen →

Sized at the format most people actually download, not at FP16. The catalogue ordered by downloads →

Find the AMD Ryzen AI Max+ 395 with Radeon 8060S: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

The Ryzen AI Max+ 395, model by model

One page per model: whether it fits this device, at which formats, and how fast.

Under the hood

The specification behind every figure

What the manufacturer publishes for this device, and the pages it was read from.

Manufacturer specification

Memory scopeshared-system
Capacity128 GB
Published options64 GB, 128 GB
Bandwidth256 GB/s
Memory typeLPDDR5x system memory
Bus widthNot published
FP32 peakNot published
Dense matrix peakNot published without sparsity
Default TDP55 W

Source ledger

AMD Ryzen AI Halo Developer Platform — Ryzen AI Max+ 395 ↗
Retrieved 2026-08-30
AMD Ryzen AI Max+ 395 Processor ↗
Retrieved 2026-08-30
Framework Desktop — configurations ↗
Retrieved 2026-09-13

Caveats

  • 256GB/s is system-memory bandwidth available to the integrated GPU, not dedicated VRAM bandwidth. The 128GB value is maximum supported system memory, not VRAM.
  • No dense matrix throughput is catalogued for this device: AMD publishes NPU TOPS for this part, not a matrix throughput for its integrated GPU. Time to first token is withheld rather than priced against a sparsity figure.
  • AMD publishes one bandwidth (256GB/s) for the platform, not per memory size; it is applied to the 64GB configuration as published, which is the same LPDDR5x-8000 memory on a smaller module set.

Machines built on the AMD Ryzen AI Max+ 395 with Radeon 8060S

HP Z2 Mini G1a

Sold with 128 GB · up to 96 GB assignable to the GPU

  • memory and GPU assignment: “Memory Up to 128 GB Unified Memory The unified memory architecture allows you to assign up to 96 GB of memory” — www.hp.com, read 2026-09-13
  • HP sells processor and memory configurations below the maximum; only the 128 GB maximum is recorded because the smaller options were not read from HP.
  • HP states up to 96 GB can be assigned to graphics. Use 96 GB as usable memory in the calculator for this machine: the installed 128 GB is not available to the model in full.
  • The processor in a given configuration may be a different Ryzen AI Max part than the 395; confirm the part before relying on the bandwidth figure.

Size a model on the 128 GB HP Z2 Mini G1a with 96 GB usable →

Framework Desktop (Ryzen AI Max+ 395)

Sold with 64 / 128 GB · up to 96 GB assignable to the GPU in the 128 GB model

  • configurations: “Max+ 395 - 64GB CPU: 3.0GHz base clock Up to 5.1GHz max boost 16-core/32-thread 64MB L3 Cache iGPU: Radeon™ 8060S Graphics Memory: 64GB LPDDR5x-8000 … Max+ 395 - 128GB … iGPU: Radeon™ 8060S Graphics Memory: 128GB LPDDR5x-8000” — frame.work, read 2026-09-13
  • GPU-addressable memory: “up to 96GB of graphics addressable memory and a 256-bit memory bus” — frame.work, read 2026-09-13
  • Framework states up to 96GB of graphics addressable memory; 96 GB is recorded as usable memory for the 128GB configuration. The page does not state the figure for the 64GB configuration, so it is not recorded for it.
  • Framework also sells a Ryzen AI Max 385 with 32GB (Radeon 8050S). That is a different processor, not catalogued, and the 395 figures are not assigned to it.
  • A 192GB option is listed as "coming soon"; it is not recorded until it is sold.
  • Read in a browser on 2026-09-13; frame.work refuses automated retrieval.

Size a model on the 64 GB Framework Desktop (Ryzen AI Max+ 395) →Size a model on the 128 GB Framework Desktop (Ryzen AI Max+ 395) with 96 GB usable →

The smaller configurations are sized on their installed memory: the maker states no GPU-assignable limit for them, so none is assumed.

Published capacity is a hardware ceiling, not guaranteed free runtime memory. The calculator shows the runtime reserve separately rather than folding it into a single number.

Run the diagnostic on the AMD Ryzen AI Max+ 395 with Radeon 8060S →
llmbottleneck
catalogue 2026-10-03models 327devices 135