No — not on this device

Ornith 1.5 397B-A17B at Q4_K_M needs 224.1 GB against 44.8 GB usable, and the shortfall of 179.3 GB is more than 32 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.

Does not fit 8K context Q4_K_M · 223.2 GB MIT Released 19 Aug 2026

The flagship Ornith on the Qwen3.5 397B-A17B architecture, MIT-licensed. A 256 GB Mac Studio at Q4, and it is in the Ollama library.

Hardware requirements What hardware do I need to run Ornith 1.5 397B-A17B? →

At Q4_K_M with an 8K context: a 512 GB Mac (M3 Ultra). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

Fits instead: Ornith 1.5 35B-A3B (20.9 GB) · Ornith 1.5 9B (6.1 GB)

The VRAM budget

weights 223.2 GB
Weights 223.2 GB KV cache @ 8K 0.23 GB Runtime overhead 0.6 GB Over budget 179.3 GB past 44.8 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 392.8 GB 393.7 GB — ~0.9 −0.1% ppl 348.9 GB over
Q6_K 303.2 GB 304.0 GB — ~1.2 −0.4% ppl 259.2 GB over
Q5_K_M 262.0 GB 262.9 GB — ~1.5 −0.8% ppl 218.1 GB over
Q4_K_M 223.2 GB 224.1 GB — ~1.8 −1.9% ppl 179.3 GB over
Q3_K_M 180.7 GB 181.5 GB — ~2.3 −5.4% ppl 136.7 GB over
Q2_K 154.8 GB 155.7 GB — ~2.9 −15% ppl 110.9 GB over

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. Only 15 of its 60 blocks keep a per-token KV cache; the rest are linear-attention, Mamba or convolution blocks with a fixed-size state.

How to run it

terminal
$ llama-server \
    -hf ornith-ai/Ornith-1.5-397B:Q4_K_M \
    -c 8192 -ngl 11

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 223.2 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02llama.cpp and Ollama spread the layers over both cards on their own. The cards take turns on every token, so expect about one card's speed; vLLM's --tensor-parallel-size 2 runs them in parallel instead.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on a 2× GeForce RTX 3090

ModelTotalTok/s est.
Ornith 1.5 35B-A3B
35.9B MoE · same family
Q4_K_M · 20.9 GB 117 Runs great
Ornith 1.5 9B
9.41B · same family
Q4_K_M · 6.1 GB 107 Runs great
Qwen3.5 397B-A17B
403B MoE · similar size
Q4_K_M · 227.4 GB — Won't fit
GLM-5.3-Flash 320B-A18B
321B MoE · similar size
Q4_K_M · 181.2 GB — Won't fit
DeepSeek V4.1 Flash 552B-A16B
552B MoE · similar size
Q4_K_M · 311.0 GB — Won't fit
DeepSeek V4 Flash 284B-A13B
284B MoE · similar size
Q4_K_M · 160.7 GB — Won't fit
Qwen3 235B-A22B
235B MoE · similar size
Q4_K_M · 134.2 GB — Won't fit

Every model against the 2× GeForce RTX 3090 →

Ornith 1.5 397B-A17B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
224.1 GB — Won't fit
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
224.1 GB — Won't fit
M4 · 16 GB
10.7 GB usable · 120 GB/s
224.1 GB — Won't fit
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
224.1 GB — Won't fit
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
224.1 GB — Won't fit
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
224.1 GB — Won't fit
M3 Ultra · 512 GB
384.0 GB usable · 800 GB/s
224.1 GB 22 Runs great

All 114 devices, by tier →