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. 1666 of 2118 indexed models fit at 8K 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 1442vision language 124embedding 26image 2video 12audio asr 39audio tts 21

What fits at 8K context

largest quantization that fits, per model · 1666 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
UncensoredLM-DeepSeek-R1-Distill-Qwen-14BQ4_K_S14.2B7.69 GiB0.76 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%
Muse-Glimmer-30BIQ2_XXS29.8B8.31 GiB0.10 GiB9.30 GiB0.00 GiB24±30%
GLM-4.7-Flash-DerestrictedMoEI1-IQ2_XS31.2B8.26 GiB0.22 GiB9.29 GiB0.01 GiB84±37%
Huihui-GLM-4.7-Flash-abliteratedMoEI1-IQ2_XS31.2B8.26 GiB0.22 GiB9.29 GiB0.01 GiB84±37%
Nemotron-3-Embed-8B-BF16Q8_08.0B7.88 GiB0.56 GiB9.28 GiB0.02 GiB24±30%
Llama3.2-30B-A3B-II-Dark-Champion-INSTRUCT-Heretic-Abliterated-UncensoredMoEI1-IQ2_XXS30.0B7.69 GiB0.78 GiB9.28 GiB0.02 GiB51±37%
v6-Finch-7B-HFQ6_K_L7.6B6.31 GiB2.13 GiB9.28 GiB0.02 GiB24±30%
rwkv-6-world-7bQ6_K_L7.6B6.31 GiB2.13 GiB9.28 GiB0.02 GiB24±30%
Skywork-R1V3-38BIQ2_XXS38.4B8.41 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-ThinkingI1-IQ2_S23.4B7.08 GiB1.34 GiB9.27 GiB0.03 GiB24±30%
HunyuanImage-2.1Q3_K_S17.5B8.42 GiB0.00 GiB9.27 GiB0.03 GiB24±30%
Qwen2.5-14B-Instruct-abliterated-v2IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
Qwen2.5-14B-Instruct-UncensoredIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
Qwen2.5-14B-Instruct-1M-abliteratedIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
Ektome-Qwen2.5-Coder-14B-Instruct-PristinelyUncensoredIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
14B-Qwen2.5-Kunou-v1IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
FinetunedQwen14BIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
DeepSeek-R1-Distill-Qwen-14B-abliterated-v2IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
C1-TachuIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
DeepSeek-R1-Distill-Qwen-14B-abliteratedIQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
Tessera-4IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
Tessera-4.1IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
EVA-Qwen2.5-14B-v0.2IQ4_XS14.8B7.62 GiB0.80 GiB9.27 GiB0.03 GiB24±30%
DeepSeek-R1-Distill-Qwen-14B-UncensoredIQ4_XS14.8B7.62 GiB0.80 GiB9.26 GiB0.04 GiB24±30%
Magistry-24B-v1.1IQ2_S23.6B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
GLM-4.7-Flash-REAP-23B-A3BMoEQ2_K_L23.0B8.24 GiB0.22 GiB9.26 GiB0.04 GiB76±37%
Gemma-4-31B-Isometry-RPI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Gemma-4-Dark-Gemistry-31BI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Prosopon-31BI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Gemma-4-Novelist-Eclipse-31BI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Giftige-Blume-31B-v1-StyleSwapI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
G4-MeroMero-31B-StyleSwapI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Gemma-4-31B-StyleTune-heretic-araI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Pantheon-Reasoning-31B-1.1I1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Gemma-4-31B-StyleTuneI1-IQ1_S32.7B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
Barcenas-StyleTune-31B-FableI1-IQ1_S32.1B7.10 GiB1.29 GiB9.26 GiB0.04 GiB24±30%
gemma-7bI1-Q6_K8.5B6.53 GiB1.86 GiB9.26 GiB0.04 GiB24±30%
Devstral-Small-2-24B-Instruct-2512UD-IQ2_M24.0B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Qwen3-VL-30B-A3B-ThinkingMoEIQ2_XS31.1B8.07 GiB0.40 GiB9.26 GiB0.04 GiB76±37%
MiroThinker-v1.0-30BMoEIQ2_XS30.5B8.07 GiB0.40 GiB9.26 GiB0.04 GiB76±37%
Qwen3-30B-A3B-Instruct-2507MoEIQ2_XS30.5B8.07 GiB0.40 GiB9.26 GiB0.04 GiB76±37%
Qwen3-30B-A3B-Thinking-2507MoEIQ2_XS30.5B8.07 GiB0.40 GiB9.26 GiB0.04 GiB76±37%
Mistral-Small-3.2-24B-Instruct-2506UD-IQ2_M24.0B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Devstral-Small-2507UD-IQ2_M23.6B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Devstral-Small-2505UD-IQ2_M23.6B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Magistral-Small-2509UD-IQ2_M24.0B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Magistral-Small-2507UD-IQ2_M23.6B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Mistral-Small-3.1-24B-Instruct-2503UD-IQ2_M24.0B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Magistral-Small-2506UD-IQ2_M23.6B7.68 GiB0.66 GiB9.26 GiB0.04 GiB24±30%
Tongyi-DeepResearch-30B-A3BMoEIQ2_XS30.5B8.07 GiB0.40 GiB9.26 GiB0.04 GiB76±37%
DeepSeek-Coder-V2-Lite-BaseMoEI1-Q4_015.7B8.32 GiB0.13 GiB9.25 GiB0.05 GiB76±37%
Wan2.1-T2V-14BQ4_014.3B8.41 GiB0.00 GiB9.25 GiB0.05 GiB24±30%
Phi-3-mini-4k-instructKV unresolvedQ3_K_L3.8B6.84 GiB1.59 GiB9.24 GiB0.06 GiB24±30%
Tini-Cybersec-8B-A1BMoEQ8_08.5B8.39 GiB0.05 GiB9.24 GiB0.06 GiB68±37%
LFM2.5-8B-A1B-KO-SFTMoEQ8_08.5B8.39 GiB0.05 GiB9.24 GiB0.06 GiB68±37%
LFM2.5-8B-A1B-SOMPOA-heresyMoEQ8_08.5B8.39 GiB0.05 GiB9.24 GiB0.06 GiB68±37%
Huihui-LFM2.5-8B-A1B-abliteratedMoEQ8_08.5B8.39 GiB0.05 GiB9.24 GiB0.06 GiB68±37%
LFM2.5-8B-A1BMoEQ8_08.5B8.39 GiB0.05 GiB9.24 GiB0.06 GiB68±37%
●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?
1666 of 2118 indexed open-weight models fit a Arc B570 10GB at 8,192 context with q8_0 KV cache, the largest being UncensoredLM-DeepSeek-R1-Distill-Qwen-14B at Q4_K_S. 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.