AMD · unified x86

AMD Ryzen AI Max+ 395 (Radeon 8060S)

AMD Ryzen AI Max+ 395 (Radeon 8060S) has 128 GB of VRAM at 256 GB/s — about 89.28 GiB usable after driver and compositor overhead. 2089 of 2118 indexed models fit at 8K context with q8_0 KV. Note only 96 GB of its 128 GB is allocatable to the GPU.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
128 GB
LPDDR5X-8000
Bandwidth
256 GB/s
256-bit bus
Tensor FP16
—
dense
TDP
120 W
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 1794vision language 191audio tts 21image 2audio asr 39video 16embedding 26

What fits at 8K context

largest quantization that fits, per model · 2089 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
Behemoth-X-123B-v2Q6_K123B93.68 GiB1.46 GiB95.99 GiB0.01 GiB1±25%
Mistral-Large-Instruct-2411Q6_K123B93.68 GiB1.46 GiB95.99 GiB0.01 GiB1±25%
Qwen3-VL-235B-A22B-ThinkingMoEQ3_K_S236B94.48 GiB0.78 GiB95.99 GiB0.01 GiB7±37%
Qwen3-VL-235B-A22B-InstructMoEQ3_K_S236B94.48 GiB0.78 GiB95.99 GiB0.01 GiB7±37%
Qwen3-235B-A22BMoEQ3_K_S235B94.48 GiB0.78 GiB95.99 GiB0.01 GiB7±37%
Qwen3-235B-A22B-abliteratedMoEI1-Q3_K_S235B94.48 GiB0.78 GiB95.99 GiB0.01 GiB7±37%
Laguna-S-2.1MoEQ6_K_L118B94.83 GiB0.27 GiB95.83 GiB0.17 GiB8±37%
MiMo-V2-FlashMoEKV unresolvedUD-IQ2_XXS310B94.58 GiB0.50 GiB95.83 GiB0.17 GiB9±37%
DeepSeek-Coder-V2-Instruct-0724MoEIQ3_S236B94.70 GiB0.28 GiB95.72 GiB0.28 GiB9±37%
DeepSeek-V2.5MoEIQ3_S236B94.70 GiB0.28 GiB95.72 GiB0.28 GiB9±37%
DeepSeek-Coder-V2-InstructMoEIQ3_S236B94.70 GiB0.28 GiB95.72 GiB0.28 GiB9±37%
Step-3.5-Flash-REAP-121B-A11BI1-Q6_K121B92.51 GiB2.14 GiB95.38 GiB0.62 GiB2±25%
grok-2MoEQ2_K_L270B93.41 GiB1.06 GiB95.32 GiB0.68 GiB3±37%
MiniMax-M3MoEIQ1_M427B93.82 GiB0.50 GiB95.03 GiB0.97 GiB9±37%
gemma-4-26B-A4B-it-Uncensored-MAXMoEF3225.8B94.02 GiB0.32 GiB95.02 GiB0.98 GiB2±25%
MiniMax-M2.1MoEI1-IQ3_M229B93.13 GiB1.03 GiB94.84 GiB1.16 GiB9±37%
MiniMax-M2.5MoEI1-IQ3_M229B93.13 GiB1.03 GiB94.84 GiB1.16 GiB9±37%
Mixtral-8x22B-Instruct-v0.1MoEQ5_K_M141B93.11 GiB0.93 GiB94.80 GiB1.20 GiB3±37%
Mixtral-8x22B-v0.1MoEQ5_K_M141B93.11 GiB0.93 GiB94.80 GiB1.20 GiB3±37%
dots.llm1.instMoEQ4_K_L143B89.95 GiB4.12 GiB94.79 GiB1.21 GiB6±37%
Mixtral-8x22B-v0.1MoEQ5_K_M141B93.10 GiB0.93 GiB94.79 GiB1.21 GiB3±37%
GLM-4.6-REAP-268B-A32BMoEQ2_K_L269B92.11 GiB1.53 GiB94.38 GiB1.62 GiB6±37%
Qwen3.5-122B-A10B-hereticMoEI1-Q6_K123B93.42 GiB0.10 GiB94.25 GiB1.75 GiB9±37%
MiniMax-M2.7MoEQ3_K_S229B92.49 GiB1.03 GiB94.21 GiB1.79 GiB9±37%
Llama-4-Maverick-17B-128E-InstructMoEKV unresolvedTQ1_0402B92.59 GiB0.80 GiB94.12 GiB1.88 GiB11±37%
MiniMax-M2MoEQ3_K_S229B92.31 GiB1.03 GiB94.02 GiB1.98 GiB9±37%
GLM-4.5-Air-DerestrictedMoEQ6_K110B92.37 GiB0.76 GiB93.86 GiB2.14 GiB7±37%
GLM-4.5-AirMoEQ6_K110B92.37 GiB0.76 GiB93.86 GiB2.14 GiB7±37%
MiniMax-M2.1-REAP-139B-A10BMoEI1-Q5_K_M139B91.98 GiB1.03 GiB93.69 GiB2.31 GiB7±37%
m51Lab-MiniMax-M2.7-REAP-139B-A10BMoEI1-Q5_K_M139B91.98 GiB1.03 GiB93.69 GiB2.31 GiB7±37%
MiniMax-M2.7-BF16-ultra-uncensored-hereticMoEI1-IQ3_S229B91.94 GiB1.03 GiB93.65 GiB2.35 GiB9±37%
DeepSeek-V4-FlashMoEQ2_K291B92.86 GiB0.03 GiB93.64 GiB2.36 GiB10±37%
Step-3.7-FlashUD-IQ4_NL201B90.63 GiB2.14 GiB93.50 GiB2.50 GiB2±25%
Mistral-Small-4-119B-2603MoEUD-Q6_K119B92.60 GiB0.09 GiB93.43 GiB2.57 GiB10±37%
GLM-4.7MoEUD-IQ1_S358B90.50 GiB1.53 GiB92.77 GiB3.23 GiB7±37%
GLM-4.7-REAP-218B-A32BMoEQ3_K_S218B90.39 GiB1.53 GiB92.65 GiB3.35 GiB6±37%
GLM-4.5MoEUD-IQ1_S358B90.38 GiB1.53 GiB92.65 GiB3.35 GiB7±37%
GLM-4.6MoEUD-IQ1_S357B90.28 GiB1.53 GiB92.55 GiB3.45 GiB7±37%
Qwen3.5-397B-A17BMoEIQ1_M403B91.53 GiB0.12 GiB92.41 GiB3.59 GiB11±37%
Kimi-Linear-48B-A3B-InstructMoEBF1649.1B91.54 GiB0.13 GiB92.37 GiB3.63 GiB2±25%
command-a-plus-05-2026-bf16MoEQ3_K_S219B91.28 GiB0.36 GiB92.35 GiB3.65 GiB7±37%
Qwen3-235B-A22B-Instruct-2507MoEIQ3_XS235B90.25 GiB0.78 GiB91.76 GiB4.24 GiB7±37%
Qwen3-235B-A22B-Thinking-2507MoEIQ3_XS235B90.25 GiB0.78 GiB91.76 GiB4.24 GiB7±37%
Solar-Open2-250BMoEIQ3_XXS250B89.86 GiB0.80 GiB91.38 GiB4.62 GiB10±37%
NVIDIA-Nemotron-3-Super-120B-A12B-BF16MoEQ5_K_L124B90.28 GiB0.37 GiB91.34 GiB4.66 GiB9±37%
Hermes-4-405BUD-IQ1_M406B88.23 GiB2.09 GiB91.30 GiB4.70 GiB2±25%
MiMo-V2.5MoEKV unresolvedUD-IQ2_M311B89.93 GiB0.50 GiB91.17 GiB4.83 GiB9±37%
GLM-4.6VMoEQ6_K_L108B89.58 GiB0.76 GiB91.08 GiB4.92 GiB7±37%
Nex-N2-ProMoEIQ1_M397B90.02 GiB0.12 GiB90.89 GiB5.11 GiB11±37%
GLM-4.5VMoEI1-Q6_K108B89.28 GiB0.76 GiB90.77 GiB5.23 GiB7±37%
Hermes-3-Llama-3.1-405BIQ1_M406B87.08 GiB2.09 GiB90.15 GiB5.85 GiB2±25%
Trinity-Large-PreviewMoEIQ2_XXS399B88.58 GiB0.67 GiB89.97 GiB6.03 GiB11±37%
Trinity-Large-TrueBaseMoEIQ2_XXS399B88.58 GiB0.67 GiB89.97 GiB6.03 GiB11±37%
gpt-oss-20b-hereticMoEQ5_120.9B88.57 GiB0.11 GiB89.36 GiB6.64 GiB5±37%
Huihui-gpt-oss-20b-BF16-abliteratedMoEQ5_120.9B88.57 GiB0.11 GiB89.36 GiB6.64 GiB5±37%
Dolphin3.0-R1-Mistral-24BF3223.6B87.82 GiB0.66 GiB89.30 GiB6.70 GiB2±25%
Dolphin3.0-Mistral-24BF3223.6B87.82 GiB0.66 GiB89.30 GiB6.70 GiB2±25%
Mistral-Small-24B-Instruct-2501-abliteratedF3223.6B87.82 GiB0.66 GiB89.30 GiB6.70 GiB2±25%
Mistral-Small-24B-Instruct-2501F3223.6B87.82 GiB0.66 GiB89.30 GiB6.70 GiB2±25%
Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTPQ6_K27.8B88.00 GiB0.27 GiB89.03 GiB6.97 GiB2±25%
●From the file◐Predictedwhat these mean

Speed is modeled, not measured: decode is memory-bandwidth bound, so tokens per second is bytes read per token against achievable bandwidth. Mixture-of-experts models carry a wider band because only the routed experts are read each step, and few have been measured publicly.

Questions people ask

What AI models can a AMD Ryzen AI Max+ 395 (Radeon 8060S) run?
2089 of 2118 indexed open-weight models fit a AMD Ryzen AI Max+ 395 (Radeon 8060S) at 8,192 context with q8_0 KV cache, the largest being Behemoth-X-123B-v2 at Q6_K. That covers text, vision-language, image, video and speech models.
How much usable memory does a AMD Ryzen AI Max+ 395 (Radeon 8060S) actually have?
Its nameplate is 128 GB, but about 89.28 GiB is available to a model once driver and compositor overhead is accounted for, and only 96 GB of the pool can be allocated to the GPU at all.
Is a AMD Ryzen AI Max+ 395 (Radeon 8060S) fast for local AI?
Its memory bandwidth is 256 GB/s, and that figure — not teraflops — is what governs token generation speed. Capacity decides what you can run; bandwidth decides how fast it runs.