DavidAU · text

Llama-3.2-4X3B-MOE-Hell-California-10B

DavidAU/Llama-3.2-4X3B-MOE-Hell-California-10B

Llama-3.2-4X3B-MOE-Hell-California-10B at Q4_K_M is exactly 6,082,137,088 bytes (5.66 GiB / 6.08 GB) — an effective 4.891 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)
Parameters
9.9B
Architecture
llama
Context
—
native (config.json)
License
apache-2.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
Q2_K3.53 GiB3,790,191,6163.048—DavidAU
Q3_K_S4.13 GiB4,438,512,6403.569—DavidAU
Q3_K_M4.54 GiB4,873,016,3203.918—DavidAU
Q3_K_L4.90 GiB5,256,008,7044.226—DavidAU
IQ4_XS5.08 GiB5,456,063,4884.387—DavidAU
Q4_K_S5.35 GiB5,746,772,9924.621—DavidAU
Q4_K_M5.66 GiB6,082,137,0884.891—DavidAU
Q5_K_S6.43 GiB6,901,746,6885.550—DavidAU
Q5_K_M6.61 GiB7,094,766,5925.705—DavidAU
Q6_K7.61 GiB8,170,685,4406.570—DavidAU
Q8_09.85 GiB10,580,055,0408.508—DavidAU

Compare with

same modality, comparable size
vs Qwen3-Coder-30B-A3B-Instructvs Qwen3.6-27Bvs Qwen3.8-27Bvs gemma-4-12B-it-qat-q4_0-unquantizedvs gemma-4-E4B-itvs Qwen3-30B-A3B-Thinking-2507All alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6500 XT 4GBGeForce RTX 3050 6GBGeForce RTX 5050 8GBGeForce RTX 3080 10GBGeForce RTX 2080 Ti 11GBGeForce RTX 5070 12GBGeForce RTX 5060 Ti 16GBApple M3 Pro 18GBGeForce RTX 3080 Ti 20GBGeForce RTX 5090 D V2 24GBGeForce RTX 5090 D 32GBApple M5 Max 36GBApple M5 Pro 48GBApple M5 Max 64GBApple M3 Ultra 96GBApple M5 Max 128GBApple M2 Ultra 192GBApple M3 Ultra 256GBApple M3 Ultra 512GB

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 5.21 GiB. The real file is 5.66 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

Architecture unavailable — this repository is gated and no ungated mirror was found. Exact file sizes above are still authoritative; only the KV math needs the config.

Questions people ask

How much VRAM does Llama-3.2-4X3B-MOE-Hell-California-10B need?
Q4_K_M is exactly 6,082,137,088 bytes (5.66 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of Llama-3.2-4X3B-MOE-Hell-California-10B should I use?
Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.