Qwen3-Zero-Coder-Reasoning-V2-0.8B · mradermacher

Qwen3-Zero-Coder-Reasoning-V2-0.8B I1-IQ2_XXS

This file is exactly 287,830,336 bytes — 0.27 GiB / 0.29 GB at an effective 2.821 bits per weight. The nominal rate for I1-IQ2_XXS is lower; mixed-precision tensors make the real figure higher, always.

From the file· 1 file(s), summed

Get it

0.27 GiB · 1 file
llama.cpp
llama-cli -hf mradermacher/Qwen3-Zero-Coder-Reasoning-V2-0.8B-i1-GGUF:I1-IQ2_XXS

Downloads and runs in one step, resolving the quantization by name.

Hugging Face CLI
hf download mradermacher/Qwen3-Zero-Coder-Reasoning-V2-0.8B-i1-GGUF Qwen3-Zero-Coder-Reasoning-V2-0.8B.i1-IQ2_XXS.gguf
Direct download

Straight from the Hugging Face CDN — we host nothing and earn nothing from this. Verify what you received against the exact byte count above; a size mismatch is the usual cause of a file that will not load.

Repository →
Size
0.27 GiB
0.29 GB
Effective bpw
2.821
from real bytes ÷ params
Tensors
—
Header
—
GGUF metadata

Fit by accelerator

AcceleratorMemory○Total @4K◐Total @32K◐Fits◐tok/s @4K◐
Arc A310 4GB4 GB1.70 GiB6.29 GiByes4K only65±30%
Arc A350M 4GB4 GB1.70 GiB6.29 GiByes4K only60±30%
Arc A370M 4GB4 GB1.70 GiB6.29 GiByes4K only60±30%
Arc A530M 4GB4 GB1.70 GiB6.29 GiByes4K only107±30%
Arc Pro A30M 4GB4 GB1.70 GiB6.29 GiByes4K only60±30%
Radeon Pro W64004 GB1.80 GiB6.39 GiByes4K only80±26.5%
Radeon RX 64004 GB1.80 GiB6.39 GiByes4K only80±26.5%
Radeon RX 6500 XT4 GB1.80 GiB6.39 GiByes4K only88±26.5%
RTX A4004 GB1.90 GiB6.49 GiByes4K only58±22%
Arc A380 6GB6 GB1.70 GiB6.29 GiByes4K only92±30%
Arc Pro A40 6GB6 GB1.70 GiB6.29 GiByes4K only94±30%
Arc Pro A50 6GB6 GB1.70 GiB6.29 GiByes4K only94±30%
GeForce RTX 20606 GB1.70 GiB6.29 GiByes4K only223±12.9%
GeForce RTX 30506 GB1.70 GiB6.29 GiByes4K only122±12.9%
GeForce RTX 3060 OEM6 GB1.70 GiB6.29 GiByes4K only223±12.9%
RTX A20006 GB1.90 GiB6.49 GiByes4K only162±22%
Apple M18 GB1.45 GiB6.04 GiByes4K only46±8.3%
Apple M28 GB1.45 GiB6.04 GiByes4K only60±8.3%
Apple M38 GB1.45 GiB6.04 GiByes4K only60±8.3%
Apple M48 GB1.45 GiB6.04 GiByes4K only66±8.3%
Arc A530M 8GB8 GB1.70 GiB6.29 GiByes107±30%
Arc A550M 8GB8 GB1.70 GiB6.29 GiByes107±30%
Arc A570M 8GB8 GB1.70 GiB6.29 GiByes107±30%
Arc A580 8GB8 GB1.70 GiB6.29 GiByes191±30%
○Spec sheet◐Predictedwhat these mean

Memory at context

ContextWeights●KV (f16)●KV (q8_0)●Working buffer◐Total (f16)◐
4,0960.27 GiB0.66 GiB0.35 GiB0.27 GiB1.20 GiB
8,1920.27 GiB1.31 GiB0.70 GiB0.27 GiB1.85 GiB
16,3840.27 GiB2.63 GiB1.39 GiB0.27 GiB3.17 GiB
32,7680.27 GiB5.25 GiB2.79 GiB0.27 GiB5.79 GiB
65,5360.27 GiB10.50 GiB5.58 GiB0.27 GiB11.04 GiB
131,0720.27 GiB21.00 GiB11.16 GiB0.27 GiB21.54 GiB
●From the file◐Predictedwhat these mean

Quantizing the KV cache to q8_0 is a roughly 2× lever on the dominant term at long context, and it is the single most useful setting most local users never touch. Totals here exclude the allocator reserve your driver takes, which is hardware-specific — the per-accelerator table above includes it.