Intel · consumer

Arc B570 10GB

Arc B570 10GB has 10 GB of VRAM at 380 GB/s — about 9.30 GiB usable after driver and compositor overhead. 1682 of 2118 indexed models fit at 4K context with q8_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
10 GB
GDDR6
Bandwidth
380 GB/s
160-bit bus
Tensor FP16
—
dense
TDP
150 W
$219 MSRP
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 1457audio asr 39vision language 125image 2video 12audio tts 21embedding 26

What fits at 4K context

largest quantization that fits, per model · 1682 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
Aurora-Code-1MoEI1-IQ2_XS34.7B8.45 GiB0.04 GiB9.30 GiB0.00 GiB116±37%
OmniAtlas-Qwen3-30B-A3BI1-IQ2_XS31.7B8.45 GiB0.00 GiB9.30 GiB0.00 GiB24±30%
Qwen3-Omni-30B-A3B-CaptionerI1-IQ2_XS31.7B8.45 GiB0.00 GiB9.30 GiB0.00 GiB24±30%
Voxtral-Mini-4B-Realtime-2602KV unresolvedF164.4B8.27 GiB0.22 GiB9.30 GiB0.00 GiB24±30%
TildeOpen-30B-Instruct-LVI1-IQ2_XXS30.7B7.91 GiB0.50 GiB9.30 GiB0.00 GiB24±30%
Llama3.2-24B-A3B-II-Dark-Champion-INSTRUCT-Heretic-Abliterated-UncensoredMoEI1-Q3_K_M18.0B8.25 GiB0.23 GiB9.29 GiB0.01 GiB62±37%
Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODERMoEI1-IQ1_S42.4B8.21 GiB0.28 GiB9.28 GiB0.02 GiB81±37%
GLM-4.7-Flash-REAP-23B-A3B-absolute-heresyMoEI1-IQ3_XXS23.0B8.36 GiB0.11 GiB9.28 GiB0.02 GiB80±37%
Gemma-The-Writer-N-Restless-Quill-10B-UncensoredI1-Q6_K10.0B7.67 GiB0.76 GiB9.28 GiB0.02 GiB24±30%
zeta-2.1Q8_08.3B8.17 GiB0.27 GiB9.28 GiB0.02 GiB24±30%
zeta-2Q8_08.3B8.17 GiB0.27 GiB9.28 GiB0.02 GiB24±30%
Skywork-R1V3-38BIQ2_XXS38.4B8.41 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
UncensoredLM-DeepSeek-R1-Distill-Qwen-14BQ4_K_M14.2B8.05 GiB0.38 GiB9.27 GiB0.03 GiB24±30%
NSFW_13B_sftQ3_K_L13.3B6.77 GiB1.66 GiB9.27 GiB0.03 GiB24±30%
Nemotron-Mini-4B-InstructQ4_K_S4.2B8.19 GiB0.27 GiB9.27 GiB0.03 GiB24±30%
Rocinante-XL-16B-v1I1-Q3_K_L16.1B7.97 GiB0.45 GiB9.27 GiB0.03 GiB24±30%
HunyuanImage-2.1Q3_K_S17.5B8.42 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
Muse-Glimmer-30BIQ2_XXS29.8B8.31 GiB0.08 GiB9.27 GiB0.03 GiB24±30%
HomunculusQ5_K_S12.5B8.08 GiB0.33 GiB9.26 GiB0.04 GiB24±30%
Falcon3-10B-InstructQ6_K_L10.3B8.06 GiB0.33 GiB9.26 GiB0.04 GiB24±30%
Phi-3-medium-128k-instructQ4_K_M14.0B7.98 GiB0.42 GiB9.25 GiB0.05 GiB24±30%
Phi-3-medium-4k-instructI1-Q4_K_M14.0B7.98 GiB0.42 GiB9.25 GiB0.05 GiB24±30%
granite-3.3-8b-instructQ8_08.2B8.09 GiB0.33 GiB9.25 GiB0.05 GiB24±30%
granite-3.2-8b-instructQ8_08.2B8.09 GiB0.33 GiB9.25 GiB0.05 GiB24±30%
Tiger-Gemma-12B-v3Q4_K_L12.8B8.02 GiB0.38 GiB9.25 GiB0.05 GiB24±30%
MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-ThinkingI1-Q2_K_S23.4B7.73 GiB0.67 GiB9.25 GiB0.05 GiB24±30%
glm-4-9b-chat-1mQ5_K_L9.5B7.07 GiB1.33 GiB9.25 GiB0.05 GiB24±30%
Wan2.1-T2V-14BQ4_014.3B8.41 GiB0.00 GiB9.25 GiB0.05 GiB24±30%
next-8bQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
next-ocrQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-GLM-4.7-Flash-Heretic-Uncensored-ThinkingQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Midas-FableAgent-8BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-Heretic-1.3.0Q8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-ThinkingQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-Instruct-Unredacted-MAXQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Poe-8B-GLM5-Opus4.6-Sonnet4.5-Kimi-Grok-Gemini-3-pro-preview-HERETICQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen-3-VL-8B-Instruct-hereticQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
nsfwcaption-qwen3-vl-8b-v3-safetensorsQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Huihui-Qwen3-VL-8B-Instruct-abliteratedQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-Reranker-8BQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Salience-1-9BQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-InstructQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-Instruct-Uncensored-V2Q8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
GRaPE-2-FlashQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Jan-v2-VL-highQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Jan-v2-VL-medQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
DeepSeek-R1-0528-Qwen3-8BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Parable-Qwen3-8B-Claude-Fable-5Q8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
ReasonCritic-7BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Finch-8B-KTOQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Finch-8BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
mythos-9b-unhinged-hereticQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
nsfwvision-qwen3-vl-8b-v3-safetensorsQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
MathSmith-hc-Qwen3-8BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Qwen3-VL-8B-Thinking-Unredacted-MAXQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
MiroThinker-v1.0-8BQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
mythos-9b-unhingedQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Maestro1-9BQ8_08.8B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
qwen3-8b-claude-agentic-fable5Q8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
Ektome-Qwen3-8B-PristinelyUncensoredQ8_08.2B8.11 GiB0.30 GiB9.24 GiB0.06 GiB24±30%
●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 Arc B570 10GB run?
1682 of 2118 indexed open-weight models fit a Arc B570 10GB at 4,096 context with q8_0 KV cache, the largest being Aurora-Code-1 at I1-IQ2_XS. That covers text, vision-language, image, video and speech models.
How much usable memory does a Arc B570 10GB actually have?
Its nameplate is 10 GB, but about 9.30 GiB is available to a model once driver and compositor overhead is accounted for.
Is a Arc B570 10GB fast for local AI?
Its memory bandwidth is 380 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.