Can Arc B580 12GB run gemma-4-26B-A4B-it?

Intel · 12 GB GDDR6 · 456 GB/s bandwidth · Google · 25.2B (active 3.8B) · MoE · model ctx up to 256K

No — gemma-4-26B-A4B-it at Q4 needs ≈ 15.24 GB even at 4K context, beyond the Arc B580 12GB's 12 GB GDDR6. It would only run with heavy CPU/disk offload. Decode at Q4/8K: ≈ 128.8 tok/s (estimated, batch 1). FP16 needs ≈ 52.16 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	15.24 GB	No	135.3 est.
Q4	8K default	15.62 GB	No	128.8 est.
Q4	32K	17.9 GB	No	98.1 est.
FP16	4K	51.78 GB	No	—
FP16	8K default	52.16 GB	No	—
FP16	32K	54.44 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 B580 12GB VRAM
Q4	13.86 GB	0.76 GB	1 GB	15.62 GB	12 GB
FP16	50.4 GB	0.76 GB	1 GB	52.16 GB	12 GB

Other GPUs that run gemma-4-26B-A4B-it

• B200 192GB SXM — ≈1339.7 tok/s Q4 est.
• MI300X 192GB — ≈887.6 tok/s Q4 est.
• H200 141GB SXM5 — ≈803.8 tok/s Q4 est.
• H100 80GB SXM5 — ≈561.0 tok/s Q4 est.
• Gaudi 3 128GB — ≈548.6 tok/s Q4 est.
• A100 80GB SXM4 — ≈341.5 tok/s Q4 est.
• RTX 5090 32GB — ≈300.1 tok/s Q4 est.
• RTX 4090 24GB — ≈168.8 tok/s Q4 est.

All GPUs for gemma-4-26B-A4B-it →

Other models for the Arc B580 12GB

• MiniCPM5-1B — ≈268.7 tok/s Q4 est.
• Qwen3.5-2B — ≈145.1 tok/s Q4 est.
• Qwen3-4B-Instruct-2507 — ≈72.5 tok/s Q4 est.
• Qwen3-4B-Thinking-2507 — ≈72.5 tok/s Q4 est.
• NVIDIA-Nemotron-3-Nano-4B-BF16 — ≈72.5 tok/s Q4 est.
• gemma-4-E2B-it — ≈56.9 tok/s Q4 est.
• Llama-3.2-3B-Instruct — ≈47.4 tok/s Q4 measured
• Mistral-7B-Instruct-v0.3 — ≈40.3 tok/s Q4 est.

All models on the Arc B580 12GB →