AMD · unified x86

AMD Ryzen AI Max+ 395 (Radeon 8060S)

AMD Ryzen AI Max+ 395 (Radeon 8060S) has 64 GB of VRAM at 256 GB/s — about 44.64 GiB usable after driver and compositor overhead. 1988 of 2118 indexed models fit at 128K context with q8_0 KV. Note only 48 GB of its 64 GB is allocatable to the GPU.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
64 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 1701image 2audio tts 21vision language 183audio asr 39video 16embedding 26

What fits at 128K context

largest quantization that fits, per model · 1988 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
codegeex4-all-9bQ3_K_M9.4B4.72 GiB42.50 GiB47.96 GiB0.04 GiB3±25%
glm-4-9b-chat-abliteratedQ3_K_M9.4B4.72 GiB42.50 GiB47.96 GiB0.04 GiB3±25%
glm-4-9b-chatQ3_K_M9.4B4.72 GiB42.50 GiB47.96 GiB0.04 GiB3±25%
Mistral-Small-Instruct-2409Q2_K22.2B32.27 GiB14.88 GiB47.90 GiB0.10 GiB3±25%
L3-DARKEST-PLANET-16.5BIQ4_XS16.5B28.30 GiB18.86 GiB47.90 GiB0.10 GiB3±25%
Llama-4-Scout-17B-16E-Instruct-abliterated-v2MoEKV unresolvedI1-Q2_K_S109B34.42 GiB12.75 GiB47.90 GiB0.10 GiB5±37%
Assistant_Pepe_70BQ2_K70.6B25.79 GiB21.25 GiB47.87 GiB0.13 GiB3±25%
Hermes-4-70BUD-IQ3_XXS70.6B25.76 GiB21.25 GiB47.83 GiB0.17 GiB3±25%
Llama-3.3-70B-InstructUD-IQ3_XXS70.6B25.76 GiB21.25 GiB47.83 GiB0.17 GiB3±25%
DeepSeek-R1-Distill-Llama-70BUD-IQ3_XXS70.6B25.76 GiB21.25 GiB47.83 GiB0.17 GiB3±25%
HunyuanImage-2.1Q5_017.5B47.04 GiB0.00 GiB47.79 GiB0.21 GiB3±25%
glm-4-9b-chat-1mIQ3_M9.5B4.53 GiB42.50 GiB47.77 GiB0.23 GiB3±25%
Huihui-Qwen3-Coder-Next-abliteratedMoEQ4_K_L79.7B45.44 GiB1.59 GiB47.72 GiB0.28 GiB15±37%
Ace-Step1.5Q8_0160M43.77 GiB3.21 GiB47.67 GiB0.33 GiB3±25%
Apertus-70B-Instruct-2509IQ3_XXS70.6B25.53 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Maenad-70BI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
DeepSeek-R1-Distill-Llama-70B-Uncensored-v2-Unbiased-ReasonerI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Rombos-LLM-70b-Llama-3.3I1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
L3.3-Electra-R1-70bI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
L3.3-70B-Magnum-v4-SEIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Latxa-Llama-3.1-70B-Instruct-v2I1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Llama-3.3_70_b_uncensored_continuedI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Llama-3.3-70B-Instruct-abliteratedI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
grok-oss-Revenant-70BI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Llama-3.1-Nemotron-70B-Instruct-HFI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
L3.3-70B-Euryale-v2.3I1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Hermes-3-Llama-3.1-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Hermes-4-70B-hereticI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Llama-3.1-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Anubis-70B-v1.2IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Golem-70B-v1bI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
DeepSeek-R1-Distill-Llama-70B-abliteratedI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
DeepSeek-R1-Distill-Llama-70B-hereticI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Legion-V2.1-LLaMa-70BI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Tess-R1-Limerick-Llama-3.1-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
SEMIKONG-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
functionary-medium-v3.2KV unresolvedIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Llama-3.1-WhiteRabbitNeo-2-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Infinity-Instruct-7M-Gen-Llama3_1-70BI1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
New-Dawn-Llama-3-70B-32K-v1.0I1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Meta-Llama-3-70B-Instruct-abliterated-v3.5I1-IQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Athene-70BIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
L3.3-70B-Magnum-DiamondIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
Meta-Llama-3-70B-InstructIQ3_XXS70.6B25.58 GiB21.25 GiB47.66 GiB0.34 GiB3±25%
IQuest-Coder-V1-40B-InstructI1-Q5_K_S39.8B25.53 GiB21.25 GiB47.58 GiB0.42 GiB3±25%
deepseek-llm-67b-chatI1-IQ2_M67.4B21.53 GiB25.23 GiB47.56 GiB0.44 GiB3±25%
deepseek-llm-67b-baseI1-IQ2_M67.4B21.53 GiB25.23 GiB47.56 GiB0.44 GiB3±25%
openbuddy-deepseek-67b-v15.3-4kI1-IQ2_M67.4B21.53 GiB25.23 GiB47.56 GiB0.44 GiB3±25%
Qwen2.5-Coder-32B-InstructQ3_K_M32.8B29.68 GiB17.00 GiB47.48 GiB0.52 GiB3±25%
Qwen3-Coder-Next-Opus-4.6-Reasoning-DistilledMoEQ4_K_M—45.16 GiB1.59 GiB47.44 GiB0.56 GiB15±37%
GLM-4.6VMoEUD-IQ1_S108B34.49 GiB12.22 GiB47.44 GiB0.56 GiB5±37%
CalmeRys-78B-Orpo-v0.1I1-IQ1_M78.0B23.68 GiB22.84 GiB47.35 GiB0.65 GiB3±25%
calme-2.3-rys-78bIQ1_M78.0B23.68 GiB22.84 GiB47.35 GiB0.65 GiB3±25%
c4ai-command-r-plus-08-2024IQ2_XS104B29.46 GiB17.00 GiB47.34 GiB0.66 GiB3±25%
Rombo-LLM-V3.0-Qwen-72bI1-IQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±25%
Qwen2.5-72B-Instruct-abliteratedI1-IQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±25%
Qwen2.5-72B-Instruct-abliterated-v2I1-IQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±25%
HuatuoGPT-o1-72BIQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±25%
MiroThinker-v1.0-72BI1-IQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±25%
EVA-Qwen2.5-72B-v0.2IQ2_XS72.7B25.20 GiB21.25 GiB47.28 GiB0.72 GiB3±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?
1988 of 2118 indexed open-weight models fit a AMD Ryzen AI Max+ 395 (Radeon 8060S) at 131,072 context with q8_0 KV cache, the largest being codegeex4-all-9b at Q3_K_M. 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 64 GB, but about 44.64 GiB is available to a model once driver and compositor overhead is accounted for, and only 48 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.