AEON-7 · text · mixture of experts

Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16

AEON-7/Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16

Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16 at Q4_K_M is exactly 22,528,154,464 bytes (20.98 GiB / 22.53 GB) — an effective 5.134 bits per weight, not the nominal 4. Its KV cache at 32K is 0.63 GiB.

From the file· summed from 1 file(s)From the file· KV per layer
Parameters
35.1B
total, not active
Architecture
qwen35moe
40 layers
Context
262,144
native (config.json)
License
other

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
IQ1_M7.67 GiB8,239,209,2801.877—vcruz305
Q4_K_M20.98 GiB22,528,154,4645.134753mrexodia
NVFP421.80 GiB23,402,864,9925.333—mrexodia
NVFP421.80 GiB23,402,864,9925.333—AninoDevel
Q2_K2 shards24.39 GiB26,186,777,5365.967—vcruz305
Q3_K_M2 shards31.60 GiB33,931,424,7047.732—vcruz305
Q8_035.21 GiB37,801,098,0808.614—mrexodia
Q8_035.21 GiB37,801,098,0808.614753AninoDevel
IQ4_XS2 shards35.30 GiB37,908,300,7368.638—vcruz305
Q4_K_S2 shards37.49 GiB40,256,766,9129.174—vcruz305
Q4_K_M2 shards39.94 GiB42,880,222,1449.771—vcruz305
Q5_K_M2 shards46.64 GiB50,076,664,76811.411—vcruz305
Q6_K2 shards53.76 GiB57,722,885,05613.153—vcruz305
BF1666.19 GiB71,065,943,13616.194—AninoDevel
BF1666.19 GiB71,065,943,13616.194—mrexodia
Q8_02 shards69.57 GiB74,705,290,17617.023—vcruz305

KV cache by context

computed per layer
ContextKV cache (f16)●Flat formulaOverstated byFull / windowed / recurrent
4,0960.08 GiB0.31 GiB4.00×10 / 0 / 30
8,1920.16 GiB0.63 GiB4.00×10 / 0 / 30
16,3840.31 GiB1.25 GiB4.00×10 / 0 / 30
32,7680.63 GiB2.50 GiB4.00×10 / 0 / 30
65,5361.25 GiB5.00 GiB4.00×10 / 0 / 30
131,0722.50 GiB10.00 GiB4.00×10 / 0 / 30

30 of 40 layers use linear attention, which keeps a fixed-size recurrent state instead of a per-token cache. Those layers do not grow with context at all — treating them as ordinary attention, as a flat formula does, overstates this model's cache by roughly 4.0× at long context.

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 6400 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 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 18.39 GiB. The real file is 20.98 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json
Layers
40
Attention heads
16
KV heads
2
Head dim
256
Hidden size
2048
Vocab
248,320
Sliding window
none
SWA period
—
MLA
no
Experts
256
Experts per token
8
use_sliding_window
—

Questions people ask

How much VRAM does Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16 need?
Q4_K_M is exactly 22,528,154,464 bytes (20.98 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16's KV cache?
0.63 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Is Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16 a mixture-of-experts model?
Yes — 256 experts, 8 routed per token. Every expert must be resident, but only the routed ones are read per token, which is why its memory requirement and its speed behave very differently.
Which quantization of Ornith-1.0-35B-AEON-Ultimate-Uncensored-BF16 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.