GestaltLabs · vision language

Qwen3.6-35B-A3B-NSC-ACE-SABER

GestaltLabs/Qwen3.6-35B-A3B-NSC-ACE-SABER

Qwen3.6-35B-A3B-NSC-ACE-SABER at Q4_K_M is exactly 21,166,757,664 bytes (19.71 GiB / 21.17 GB) — an effective 4.769 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)
Parameters
35.5B
Architecture
qwen35moe
Context
—
native (config.json)
License
apache-2.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
Q2_K12.05 GiB12,939,593,5042.916—GestaltLabs
Q2_K12.34 GiB13,252,289,7922.986—slevinw
Q2_K12.34 GiB13,252,289,7922.986—GestaltLabs
Q3_K_S14.14 GiB15,182,184,2243.421—GestaltLabs
Q3_K_S14.48 GiB15,551,290,6243.504—slevinw
Q3_K_S14.48 GiB15,551,290,6243.504—GestaltLabs
Q3_K_M15.61 GiB16,764,763,9363.777—GestaltLabs
Q3_K_M15.99 GiB17,170,914,5603.869—GestaltLabs
Q3_K_M15.99 GiB17,170,914,5603.869—slevinw
Q3_K_L16.87 GiB18,115,329,8244.082—GestaltLabs
Q3_K_L17.28 GiB18,556,345,6004.181—GestaltLabs
Q3_K_L17.28 GiB18,556,345,6004.181—slevinw
Q4_K_S18.52 GiB19,889,903,3924.482—GestaltLabs
Q4_K_S18.97 GiB20,370,003,2004.590—GestaltLabs
Q4_K_S18.97 GiB20,370,003,2004.590—slevinw
Q4_K_M19.71 GiB21,166,757,6644.769—GestaltLabs
Q4_K_M20.23 GiB21,716,604,1604.893—slevinw
Q4_K_M20.23 GiB21,716,604,1604.893—GestaltLabs
Q5_K_S22.33 GiB23,981,283,1045.403—GestaltLabs
Q5_K_S22.88 GiB24,565,847,2965.535—GestaltLabs
Q5_K_S22.88 GiB24,565,847,2965.535—slevinw
Q5_K_M23.03 GiB24,729,130,7845.572—GestaltLabs
Q5_K_M23.61 GiB25,349,625,0885.712—slevinw
Q5_K_M23.61 GiB25,349,625,0885.712—GestaltLabs
Q6_K26.56 GiB28,514,152,2246.425—GestaltLabs
Q6_K27.20 GiB29,209,709,8246.582—slevinw
Q6_K27.20 GiB29,209,709,8246.582—GestaltLabs
Q8_034.37 GiB36,903,139,1048.315—GestaltLabs
Q8_035.21 GiB37,801,096,4488.517—GestaltLabs
Q8_035.21 GiB37,801,096,4488.517—slevinw
F1664.61 GiB69,376,637,02415.632—GestaltLabs
F1666.19 GiB71,065,941,60016.012—GestaltLabs
F1666.19 GiB71,065,941,60016.012—slevinw

Compare with

same modality, comparable size
vs Qwen3.6-35B-A3Bvs Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTPvs Qwen3.5-9Bvs gemma-4-26B-A4B-itvs gemma-4-12B-itvs Qwen3.5-4BAll 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 18.60 GiB. The real file is 19.71 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 Qwen3.6-35B-A3B-NSC-ACE-SABER need?
Q4_K_M is exactly 21,166,757,664 bytes (19.71 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of Qwen3.6-35B-A3B-NSC-ACE-SABER 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.