Intel · workstation

Arc Pro B60 24GB

Arc Pro B60 24GB has 24 GB of VRAM at 456 GB/s — about 22.32 GiB usable after driver and compositor overhead. 1962 of 2118 indexed models fit at 16K context with q4_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
24 GB
GDDR6
Bandwidth
456 GB/s
192-bit bus
Tensor FP16
99 TF
dense
TDP
200 W
$599 MSRP
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 1685vision language 173image 2video 16audio tts 21audio asr 39embedding 26

What fits at 16K context

largest quantization that fits, per model · 1962 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16UD-Q4_K_S33.0B21.47 GiB0.00 GiB22.31 GiB0.01 GiB12±30%
Hunyuan-A13B-InstructMoEUD-TQ1_080.4B20.95 GiB0.56 GiB22.31 GiB0.01 GiB12±30%
Seed-OSS-36B-Instruct-biprojected-norm-preserving-abliteratedI1-Q4_K_M36.2B20.27 GiB1.13 GiB22.29 GiB0.03 GiB12±30%
Seed-OSS-36B-InstructQ4_K_M36.2B20.27 GiB1.13 GiB22.29 GiB0.03 GiB12±30%
Hermes-4.3-36B-hereticI1-Q4_K_M36.2B20.27 GiB1.13 GiB22.29 GiB0.03 GiB12±30%
Hermes-4.3-36BQ4_K_M36.2B20.27 GiB1.13 GiB22.29 GiB0.03 GiB12±30%
Seed-OSS-36B-BaseQ4_K_M36.2B20.27 GiB1.13 GiB22.29 GiB0.03 GiB12±30%
Qwen3.5-99BMoEI1-IQ1_M99.0B21.35 GiB0.11 GiB22.29 GiB0.03 GiB59±37%
Gemma-4-Gembrain-X-Core-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Gembrain-X-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-Isometry-Fabled-PersonaI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Versipellis-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma4-Gutenberg-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
G4-MeroMero-31B-uncensored-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Novelist-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Wanabi-Gemma4-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
G4-Alice-v1.2-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Agares-31B-v1I1-Q5_K_M30.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma4-Gutenberg-31B-HereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-Ortenzya-The-Creative-Wordsmith-31B-it-uncensored-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Gemsicle-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Gembrain-31B-it-uncensored-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Melinoe-Gemma4-31B-VL-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
G4-MeroMero-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Glistening-Gem-31B-v1.0I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Melinoe-Gemma4-31B-VLI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-Storymaxxed3I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-AssGuard-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
copywriter-gemma4-31bI1-Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-heretic-finetuneI1-Q5_K_M30.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Garnet-V2-31B-it-ultra-uncensored-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-Claude-Opus-Distill-v2Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-abliterated-v3I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Harmonia-31B-uncensored-hereticQ5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-noloopI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Webs-Sejong-31B-v7I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Lilith-31B-v1.0I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
JGOS-31B-ThinkI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-MergemaxxedI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
K1-v6-zeroI1-Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-uncensored-hereticQ5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Queen-31B-it-uncensored-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Sphinsikus-Chronist-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-hereticI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma4-31B-Finetuned-V2I1-Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-storymaxxedI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-Fable-5-Agent-DistillQ5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-uncensoredQ5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-storymaxxed2I1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-abliteratedQ5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Giftige-Blume-31B-v2Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-itQ5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-Grand-Horror-X-INTENSE-HERETIC-UNCENSORED-ThinkingI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-Mystery-Fine-Tune-HERETIC-UNCENSORED-ThinkingI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-ThinkingI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Huihui-gemma-4-31B-it-abliterated-v2I1-Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-Queen-31B-itI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
gemma-4-31b-it-heretic-araI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Monika-31BI1-Q5_K_M31.3B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±30%
Gemma-4-31B-Fable-CoderI1-Q5_K_M32.7B20.35 GiB1.03 GiB22.26 GiB0.06 GiB12±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 B60 24GB run?
1962 of 2118 indexed open-weight models fit a Arc Pro B60 24GB at 16,384 context with q4_0 KV cache, the largest being Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 at UD-Q4_K_S. That covers text, vision-language, image, video and speech models.
How much usable memory does a Arc Pro B60 24GB actually have?
Its nameplate is 24 GB, but about 22.32 GiB is available to a model once driver and compositor overhead is accounted for.
Is a Arc Pro B60 24GB fast for local AI?
Its memory bandwidth is 456 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.