← AI Hashrate

Can MI300X 192GB run Qwen3.5-27B?

AMD · 192 GB HBM3 · 5300 GB/s bandwidth · Alibaba · 27.0B · model ctx up to 256K

Yes — the MI300X 192GB (192 GB HBM3) can run Qwen3.5-27B. At Q4 with the default 8K context it needs ≈ 21.25 GB VRAM (weights 14.85 GB + KV cache 5.4 GB + 1 GB runtime overhead), which fits within 192 GB. Decode at Q4/8K: ≈ 124.9 tok/s (estimated, batch 1). FP16 (≈ 60.4 GB) also fits, at ≈ 34.4 tok/s (estimated). Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.

Fit & speed by quant and context

QuantContextVRAM neededFits?tok/s (decode)
Q44K18.55 GBYes124.9 est.
Q48K default21.25 GBYes124.9 est.
Q432K37.45 GBYes124.9 est.
FP164K57.7 GBYes34.4 est.
FP168K default60.4 GBYes34.4 est.
FP1632K76.6 GBYes34.4 est.

Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s means the model is far beyond this card (offload-only territory).

VRAM breakdown at 8K context

QuantWeightsKV cacheOverheadTotal neededMI300X 192GB VRAM
Q414.85 GB5.4 GB1 GB21.25 GB192 GB
FP1654.0 GB5.4 GB1 GB60.4 GB192 GB

Other GPUs that run Qwen3.5-27B

All GPUs for Qwen3.5-27B →

Other models for the MI300X 192GB

All models on the MI300X 192GB →