Intel · workstation

Arc Pro A40 6GB

Arc Pro A40 6GB has 6 GB of VRAM at 192 GB/s — about 5.58 GiB usable after driver and compositor overhead. 636 of 2118 indexed models fit at 64K context with q8_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
6 GB
GDDR6
Bandwidth
192 GB/s
96-bit bus
Tensor FP16
—
dense
TDP
50 W
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 491vision language 67audio asr 36audio tts 18embedding 21video 3

What fits at 64K context

largest quantization that fits, per model · 636 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
Teuken-7B-instruct-research-v0.4I1-IQ3_M7.5B3.68 GiB1.06 GiB5.58 GiB0.00 GiB22±30%
AMD-OLMo-1B-SFT-DPOIQ3_M1.2B0.53 GiB4.25 GiB5.58 GiB0.00 GiB21±30%
OpenClaude-1.7B-MergedQ4_K_M1.7B1.07 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Supertron2-Reranker-2BI1-Q4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Uni-MuMER-Qwen3-VL-2BI1-Q4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Qwen3-VL-2B-ThinkingQ4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Qwen3-VL-Reranker-2BI1-Q4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Qwen3-VL-2B-InstructQ4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
OpenCaption-2B-VL-SFT-v1.0I1-Q4_12.1B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Atomight-V2.5-1.7BI1-Q4_11.7B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
gaon-1.7b-v2-translateI1-Q4_11.7B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
gaon-1.7b-v2-instructI1-Q4_11.7B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Lightning-1.7BQ4_11.7B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
DorsetHeatwaveLLM2I1-Q4_11.7B1.06 GiB3.72 GiB5.58 GiB0.00 GiB21±30%
Qwen2.5-3B-Instruct-abliteratedI1-Q4_03.1B3.57 GiB1.20 GiB5.58 GiB0.00 GiB22±30%
canary-qwen-2.5bBF162.6B4.73 GiB0.00 GiB5.57 GiB0.01 GiB22±30%
EXAONE-Deep-7.8BQ4_K_L7.8B4.73 GiB0.00 GiB5.57 GiB0.01 GiB22±30%
EXAONE-3.5-7.8B-InstructQ4_K_L7.8B4.73 GiB0.00 GiB5.57 GiB0.01 GiB22±30%
Qwen3-1.7BQ3_K_L2.0B1.06 GiB3.72 GiB5.57 GiB0.01 GiB21±30%
Nanbeige4.1-3BQ5_K_M3.9B2.63 GiB2.13 GiB5.57 GiB0.01 GiB22±30%
Yi-6B-ChatI1-IQ3_M6.1B2.62 GiB2.13 GiB5.57 GiB0.01 GiB22±30%
Qwen3-TTS-12Hz-0.6B-BaseQ4_K_M915M4.72 GiB0.00 GiB5.57 GiB0.01 GiB22±30%
VoxCPM2F162.3B4.72 GiB0.00 GiB5.57 GiB0.01 GiB22±30%
Vero-Qwen35-9B-BaseI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Vero-Qwen35-9BI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-Claude-4.6-Opus-Deckard-V4.2-Uncensored-Heretic-ThinkingI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Morphos-9BI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwable-9B-Claude-Fable-5-hereticI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Holo-3.1-9BI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwable-9B-Claude-Fable-5I1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-imabari-v2I1-IQ3_XXS9.7B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-abliterated-v2-MAXI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
OmniCoder-9B-Claude-Opus-High-Reasoning-DistillI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwable-9B-Claude-Fable-5-StraTAI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwable-9B-Claude-Fable-5-OBLITERATEDI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-RpRMax-v1I1-IQ3_XXS9.7B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
AdQWENistrator-9BI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
cajal-9b-v2-fullI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Holo-3.1-9B-CoderI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
PlutoI1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Holo-3.1-9B-abliterated-rdoI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-BaseI1-IQ3_XXS9.7B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
qwen3.5-9b-nsfw-captioning-v5I1-IQ3_XXS9.4B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Miss_MARTHA-9B-Qwen3.5-OmniI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Qwen3.5-9B-DeepSeek-V4-FlashI1-IQ3_XXS9.7B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Huihui-Qwen3.5-9B-Claude-4.6-Opus-abliteratedI1-IQ3_XXS9.7B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
Katarau-9B-ru-RP-nsfwI1-IQ3_XXS9.0B3.67 GiB1.06 GiB5.57 GiB0.01 GiB22±30%
t5-v1_1-xxlQ2_K4.8B4.72 GiB0.00 GiB5.56 GiB0.02 GiB22±30%
Qwen3.5-4BQ6_K_L4.7B3.69 GiB1.06 GiB5.56 GiB0.02 GiB22±30%
orpheus-3b-0.1-ftUD-IQ2_XXS3.8B1.03 GiB3.72 GiB5.56 GiB0.02 GiB22±30%
SmolLM3-3BQ6_K3.1B2.36 GiB2.39 GiB5.56 GiB0.02 GiB22±30%
gemma-3n-E4B-itQ4_K_M7.8B4.23 GiB0.49 GiB5.56 GiB0.02 GiB22±30%
EXAONE-4.0-1.2B-abliteratedF161.5B2.78 GiB1.99 GiB5.55 GiB0.03 GiB22±30%
Llama-Doctor-3.2-3B-InstructI1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
Llama-3.2-3B-Instruct-roleplay-tunedI1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
Llama-3.2-3B-Instruct-heretic-ablitered-uncensoredI1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
llama-3.2-3b-instructIQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
Llama3.2-3B-creative-writer-v0.1I1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
Firefly-V3.2I1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±30%
Firefly-V3I1-IQ2_XS3.2B1.02 GiB3.72 GiB5.55 GiB0.03 GiB22±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 Pro A40 6GB run?
636 of 2118 indexed open-weight models fit a Arc Pro A40 6GB at 65,536 context with q8_0 KV cache, the largest being Teuken-7B-instruct-research-v0.4 at I1-IQ3_M. That covers text, vision-language, image, video and speech models.
How much usable memory does a Arc Pro A40 6GB actually have?
Its nameplate is 6 GB, but about 5.58 GiB is available to a model once driver and compositor overhead is accounted for.
Is a Arc Pro A40 6GB fast for local AI?
Its memory bandwidth is 192 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.