Can Arc B570 10GB run Qwen3.6-27B?

Intel · 10 GB GDDR6 · 380 GB/s bandwidth · Alibaba · 27.8B · model ctx up to 256K

No — Qwen3.6-27B at Q4 needs ≈ 19.07 GB even at 4K context, beyond the Arc B570 10GB's 10 GB GDDR6. It would only run with heavy CPU/disk offload. Decode at Q4/8K: ≈ 5.2 tok/s (estimated, batch 1). FP16 needs ≈ 62.16 GB — does not fit on this card. 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

Quant	Context	VRAM needed	Fits?	tok/s (decode)
Q4	4K	19.07 GB	No	6.8 est.
Q4	8K default	21.85 GB	No	5.2 est.
Q4	32K	38.53 GB	No	1.7 est.
FP16	4K	59.38 GB	No	—
FP16	8K default	62.16 GB	No	—
FP16	32K	78.84 GB	No	—

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

Quant	Weights	KV cache	Overhead	Total needed	Arc B570 10GB VRAM
Q4	15.29 GB	5.56 GB	1 GB	21.85 GB	10 GB
FP16	55.6 GB	5.56 GB	1 GB	62.16 GB	10 GB

Other GPUs that run Qwen3.6-27B

• B200 192GB SXM — ≈183.1 tok/s Q4 est.
• MI300X 192GB — ≈121.3 tok/s Q4 est.
• H200 141GB SXM5 — ≈109.9 tok/s Q4 est.
• H100 80GB SXM5 — ≈76.7 tok/s Q4 est.
• Gaudi 3 128GB — ≈75.0 tok/s Q4 est.
• A100 80GB SXM4 — ≈46.7 tok/s Q4 est.
• RTX 5090 32GB — ≈41.0 tok/s Q4 est.
• RTX 4090 24GB — ≈23.1 tok/s Q4 est.

All GPUs for Qwen3.6-27B →

Other models for the Arc B570 10GB

• MiniCPM5-1B — ≈223.9 tok/s Q4 est.
• Qwen3.5-2B — ≈120.9 tok/s Q4 est.
• Llama-3.2-3B-Instruct — ≈75.6 tok/s Q4 est.
• Qwen3-4B-Instruct-2507 — ≈60.5 tok/s Q4 est.
• Qwen3-4B-Thinking-2507 — ≈60.5 tok/s Q4 est.
• NVIDIA-Nemotron-3-Nano-4B-BF16 — ≈60.5 tok/s Q4 est.
• gemma-4-E2B-it — ≈47.4 tok/s Q4 est.
• Mistral-7B-Instruct-v0.3 — ≈33.6 tok/s Q4 est.

All models on the Arc B570 10GB →