Ryzen AI Max+ 395 LLM benchmarks
2,169 results · 3,570 measurements · 15 sources
Device specifications
| Variant | Memory | Memory type | Bus | Bandwidth | FP16 matrix | INT8 matrix |
|---|---|---|---|---|---|---|
| Ryzen AI Max+ 395 (Radeon 8060S) ↗ | SharedUp to 128 GB system RAM | LPDDR5x | 256-bit | — | — | — |
Manufacturer specifications. Shared system RAM is not dedicated VRAM. Unreported values remain —; measured LLM results retain their reported configuration.
Selected reported tests
24 of 2,169 resultsSource-reported results, selected across reports by documentation quality. Match configurations before comparing values.
| View ReportGeneric LLM | diffusiongemma-26B-A4B-it | JevBench231 public cases · v1.2.101 concurrent request | 81.82 %Score · 189 / 231 · Original JevBench per-case scoring | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
|---|---|---|---|---|---|---|
| View ReportGeneric LLM | diffusiongemma-26B-A4B-it | JevBench231 public cases · v1.2.101 concurrent request | 81.82 %Score · 189 / 231 · Original JevBench per-case scoring | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View ReportGeneric LLM | MinerU2.5-Pro-2605-1.2B | OmniDocBenchFull1651 pages · v1.6 | 0.04 edit distanceText edit distance · quick_match_cdm · Lower is better | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | Qwen3.6-35B-A3B | HEPAValid complete runs64K configured context | 100 %Score · n = 1 · Deterministic validity gate | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | halogen-swift1.5-qwen3.8-flash-next-v2-abliterated | WikiTextRaw test split70,774tokens1 concurrent request | 3.229 perplexityWord perplexity · Teacher-forced perplexity · Lower is better | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | Gemma4-26B-CIRU-AMD-STRIX-vLLM | GSM8KSize not reported | 95.45 %Accuracy · n = 100 | — | — | GPU count unknownUnknown · Unified memory · 128 GB RAMConfiguration
|
| View ReportCoding session | Qwen3.8-Flash-Next | Neon LadderContract v2_12 · Neon Overdrive game-generation contract v2_121 concurrent request | 16 checksScore · 16 / 19 · 19 static checks; not a human playtest grade | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportCustom evaluation | Qwen3.8-27B | HumanEval40 problems;every fourthHumanEval index · HumanEvalPlus; revision not reported1 concurrent request | 90 %pass@1 · 36 / 40 · Executable tests | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportLocalMaxxing | DeepSeek-V4-Flash | Token speed | — | 201.6 tok/sPrefill · median · per request | 30.75 tok/sGeneration · median · per request | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View Reportllama-bench | GLM-4.7-Flash | Token speed2,048 input tokens · 0 output tokens · 16,384 existing KV tokens | — | 170.42 tok/sPrefill · mean · per request | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View Reportllama-bench | gemma-4-26B-A4B-it | Token speed512 input tokens · 0 output tokens · 0 existing KV tokens | — | 909.04 tok/sPrefill · mean · per request | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View ReportGeneric LLM | halogen-qwen3.8-flash-nextConflicting evidence | Token speedladder | — | — | 73.59 tok/sGeneration · per request | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View Reportllama-bench | Llama-3.1-8B-Instruct | Token speed | 92.08 tok/sInput + output · mean · per request | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View Reportllama-bench | Qwen3-Coder-Next | Token speed | — | 496.38 tok/sPrefill · mean · per request | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View Reportllama-bench | flint | Token speed | — | 1,451 tok/sPrefill · per request | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | MinerU2.5-Pro-2605-1.2B | OmniDocBenchFull1651 pages · v1.6 | 0.12 edit distanceReading-order edit distance · quick_match_cdm · Lower is better | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | Qwen3.6-35B-A3B | HEPADeterministic checks64K configured context | 100 %Score · n = 1 · Deterministic validity gate | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | halogen-swift1.5-qwen3.8-flash-next-v2-abliterated | MMLU-ProOne question per14subjects; seed71 concurrent request | 10 correctScore · n = 14 · Accuracy | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportGeneric LLM | Gemma4-26B-CIRU-AMD-STRIX-vLLM | HumanEval100 tasks | 95.73 %pass@1 · n = 100 · Accuracy | — | — | GPU count unknownUnknown · Unified memory · 128 GB RAMConfiguration
|
| View ReportCoding session | Qwen3.8-Flash-Next | Neon LadderContract v2_12 · Neon Overdrive game-generation contract v2_121 concurrent request | 17 checksScore · 17 / 19 · 19 static checks; not a human playtest grade | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportCustom evaluation | Qwen3.8-27B | HumanEval40 problems;every fourthHumanEval index · HumanEvalPlus; revision not reported1 concurrent request | 87.5 %pass@1 · 35 / 40 · Executable tests | — | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View ReportLocalMaxxing | Qwen3.6-35B-A3B | Token speed | — | 1,467.82 tok/sPrefill · median · per request | 248.2 tok/sGeneration · median · per request | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View Reportllama-bench | GLM-4.7-Flash | Token speed0 input tokens · 128 output tokens · 16,384 existing KV tokens | — | — | 35.47 tok/sGeneration · mean · per request | 1 × Ryzen AI Max+ 395Single GPU · Unified memory · 128 GB RAMConfiguration
|
| View Reportllama-bench | gemma-4-26B-A4B-it | Token speed0 input tokens · 128 output tokens · 0 existing KV tokens | — | — | 24.64 tok/sGeneration · mean · per request | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
Models and software in this collection
Model families: DeepSeek R1, DeepSeek V4, DeepSeek V4.1, GLM 4.5, GLM 4.7, GLM 5.3, GPT-OSS, Gemma 4, Laguna, Llama 2, Llama 3.1, Llama 3.2, Ministral 3, Mistral, Muse Glimmer, Nemotron 3, Qwen-AgentWorld, Qwen2.5, Qwen3, Qwen3.5, Qwen3.6, Qwen3.8, TinyLlama
Engines: Gufo, Halogen, Loculus Core, Strata, custom, dwarfstar, halogen-flash-server, hipfire, llama.cpp, llama.cpp (Nathan Vulkan fork), llama.cpp (halo-box-strix), llama.cpp System One, lucebox, ollama, vLLM (Ciru modified), vllm
Methodology and evidence standards · Download the research dataset