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. 1197 of 2118 indexed models fit at 64K 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 1000vision language 100image 1video 12embedding 26audio asr 38audio tts 20

What fits at 64K context

largest quantization that fits, per model · 1197 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
L3-Dark-Planet-8BQ3_K_L8.0B4.21 GiB4.25 GiB9.30 GiB0.00 GiB24±30%
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%
granite-8b-code-instruct-4kI1-Q3_K_M8.1B3.67 GiB4.78 GiB9.29 GiB0.01 GiB24±30%
granite-8b-code-base-4kI1-Q3_K_M8.1B3.67 GiB4.78 GiB9.29 GiB0.01 GiB24±30%
Gemma-4-12B-StyleTuneI1-Q3_K_M13.0B6.07 GiB2.37 GiB9.29 GiB0.01 GiB24±30%
gemma-4-12b-heretic-styletune-headI1-Q3_K_M12.0B6.07 GiB2.37 GiB9.29 GiB0.01 GiB24±30%
syrian-gemma-12bI1-Q3_K_M13.0B6.07 GiB2.37 GiB9.29 GiB0.01 GiB24±30%
Apertus-8B-Instruct-2509IQ4_XS8.1B4.17 GiB4.25 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-Engineer-Deckard-GeminiI1-IQ1_M27.7B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-HERETIC-Polaris-Advanced-Thinking-Alpha-uncensoredI1-IQ1_M27.4B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-Deckard-PKD-Heretic-Uncensored-ThinkingI1-IQ1_M27.4B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Huihui-Qwen3.5-27B-abliteratedI1-IQ1_M27.8B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-Unredacted-MAXI1-IQ1_M27.4B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-hereticI1-IQ1_M27.4B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-DerestrictedI1-IQ1_M27.8B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Qwen3.5-27B-Claude-4.6-Opus-Reasoning-DistilledI1-IQ1_M27.8B6.30 GiB2.13 GiB9.29 GiB0.01 GiB24±30%
Trinity-MiniMoEIQ2_M26.1B7.90 GiB0.59 GiB9.29 GiB0.01 GiB68±37%
Teuken-7B-instruct-research-v0.4Q8_07.5B7.38 GiB1.06 GiB9.28 GiB0.02 GiB24±30%
zeta-2Q3_K_L8.3B4.19 GiB4.25 GiB9.28 GiB0.02 GiB24±30%
GLM-4.7-Flash-REAP-23B-A3BMoEUD-IQ1_S23.0B6.71 GiB1.76 GiB9.28 GiB0.02 GiB41±37%
Skywork-R1V3-38BIQ2_XXS38.4B8.41 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
legitus-instruct-v1I1-IQ4_XS8.1B4.16 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Falcon3-10B-InstructI1-IQ2_S10.3B3.09 GiB5.31 GiB9.27 GiB0.03 GiB24±30%
HunyuanImage-2.1Q3_K_S17.5B8.42 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
Foundation-Sec-8B-Instruct-hereticIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
MiniCPM-Llama3-V-2_5IQ4_XS8.5B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
llama3.1-heretic-unsensoredIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
dolphin-2.9.4-llama3.1-8bIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Dolphin3.0-Llama3.1-8B-abliteratedIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Deepseek-R1-Distill-NSFW-RPv1IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama3-ChatQA-1.5-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8B-hereticIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-ReasoningIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama3.3-8B-Instruct-Thinking-Claude-4.5-Opus-High-ReasoningIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
grok-oss-Apollyon-8B-hereticIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama3.1-GptDeluxe-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
SelfCite-8B-CC-SFTIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama-3.3-8B-Instruct-128K-JbliteratedIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
ai-girlfriend-v2IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama-3.3-8B-Instruct-128K_AbliteratedIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
grok-oss-Apollyon-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Lumen-1.2.5IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
HARC-Llama-3.1-8B-InstructIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
grok-oss-Thanatos-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
L3-8B-Sunfall-v0.5-Stheno-v3.2IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
OpenElla-NovelWriter-Requiem-AscendedIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
llama-joycaption-beta-one-hf-llavaIQ4_XS8.5B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
calme-2.3-legalkit-8bIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
DarkIdol-Llama-3.1-8B-Instruct-1.3-UncensoredIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama3.1-DilemmaDeluxe-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
DarkIdol-Llama-3.1-8B-Instruct-1.2-UncensoredIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Humanish-Roleplay-Llama-3.1-8BIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama-3.1-8B-kali-pentesterIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
NeuralDaredevil-8B-abliteratedIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Llama-3.3-8B-InstructIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Hathor_Sofit-L3-8B-v1IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
llama3-8B-DarkIdol-2.3-Uncensored-32KIQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
Hathor_Unstable-L3-8B-v0.3IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 GiB24±30%
L3-8B-Stheno-v3.2IQ4_XS8.0B4.18 GiB4.25 GiB9.27 GiB0.03 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?
1197 of 2118 indexed open-weight models fit a Arc B570 10GB at 65,536 context with q8_0 KV cache, the largest being L3-Dark-Planet-8B at Q3_K_L. 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.
Arc B570 10GB — what AI models can it run locally? — ossmodeldb