OpenAI on Intel Battlemage

Can I run gpt-oss 120B-A5.1B on an Arc B580?

Not at Q4_K_M — it needs 67.4 GB against 10.5 GB available. You would need 7 of these cards.

Won't fit

67.4 GB of 10.5 GB · 643%
070 GB
Weights 66.0 GB
KV cache 0.6 GB
Runtime overhead 0.8 GB
Over the limit 56.9 GB

Short by 56.9 GB. You can run it with 4 of 36 layers on the Arc B580 and the rest in system RAM, at roughly 11.1 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.

Generation11.1tok/s
Prompt processing1894tok/s
Max context0tokens
KV per 1K tokens0GB

Every quantisation of gpt-oss 120B-A5.1B on a Arc B580

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 217.6 GB 218.9 GB Won't fit 3.38
INT8 / W8A8 8.50 115.3 GB 116.7 GB Won't fit 6.33
FP8 (E4M3) 8.00 108.8 GB 110.1 GB Won't fit 6.85
Q6_K 6.56 89.2 GB 90.6 GB Won't fit 8.24
Q5_K_M 5.67 77.1 GB 78.5 GB Won't fit 9.65
Q5_K_S 5.52 75.1 GB 76.4 GB Won't fit 9.89
Q4_K_M 4.85 66.0 GB 67.4 GB Won't fit 11.1
Q4_K_S 4.58 62.4 GB 63.8 GB Won't fit 12.0
Q4_0 4.55 62.0 GB 63.4 GB Won't fit 12.0
AWQ 4-bit 4.25 59.4 GB 60.7 GB Won't fit 12.5
GPTQ 4-bit 4.25 59.4 GB 60.7 GB Won't fit 12.5
MXFP4 4.25 59.4 GB 60.7 GB Won't fit 12.5
IQ4_XS 4.25 58.0 GB 59.3 GB Won't fit 12.8
Q3_K_M 3.91 53.4 GB 54.7 GB Won't fit 14.1
IQ3_M 3.70 50.6 GB 51.9 GB Won't fit 14.8
IQ3_XXS 3.06 41.9 GB 43.3 GB Won't fit 17.8
Q2_K 2.63 36.1 GB 37.5 GB Won't fit 21.3
IQ2_XXS 2.06 28.5 GB 29.8 GB Won't fit 27.4
IQ1_M 1.75 24.3 GB 25.7 GB Won't fit 32.9

Also worth checking