Ornith 1.0 397B needs about 243.3 GB of memory at Q4_K_M (a 241.8 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 27 quants, from 1.1 GB to 421.6 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.0 GB.
Ornith 1.0 397B – a large‑scale, Mixture‑of‑Experts model for coding and reasoning
Ornith 1.0 397B is a 396‑billion‑parameter Mixture‑of‑Experts model based on the qwen35moe architecture. It is trained on 262144‑token context windows and is licensed under MIT. The model supports reasoning, tools, and vision tasks.
Good at
- Advanced coding tasks
- Agentic reasoning
- Vision‑enabled prompts
Limits
- Requires large VRAM for full‑precision use
- No native vision support in llama.cpp
- Limited to 262144‑token context
Best for
- Single‑GPU deployment of large coding agents
- Research on Mixture‑of‑Experts architectures
- Fine‑tuning for specialized coding benchmarks
Who it’s for. Developers and researchers who need a high‑capacity model for coding and reasoning tasks and have access to GPUs with sufficient VRAM. The MIT license allows unrestricted use and modification.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 1.1 GB | 2.6 GB | Fits | Get file → | |
| IQ1_S | 81.8 GB (3 parts) | 83.3 GB | Too big | Get file → | |
| IQ1_M | 91.4 GB (3 parts) | 92.9 GB | Too big | Get file → | |
| IQ2_XXS | 106.3 GB (3 parts) | 107.8 GB | Too big | Get file → | |
| IQ2_XS | 118.5 GB (3 parts) | 120.0 GB | Too big | Get file → | |
| IQ2_S | 120.5 GB (4 parts) | 122.0 GB | Too big | Get file → | |
| IQ2_M | 133.1 GB (4 parts) | 134.6 GB | Too big | Get file → | |
| Q2_K | 139.5 GB (4 parts) | 141.0 GB | Too big | Get file → | |
| Q2_K_L | 140.5 GB (4 parts) | 142.0 GB | Too big | Get file → | |
| IQ3_XXS | 166.1 GB (5 parts) | 167.6 GB | Too big | Get file → | |
| Q3_K_S | 172.9 GB (5 parts) | 174.4 GB | Too big | Get file → | |
| IQ3_XS | 181.4 GB (5 parts) | 183.0 GB | Too big | Get file → | |
| Q3_K_M | 181.5 GB (5 parts) | 183.0 GB | Too big | Get file → | |
| Q3_K_L | 189.5 GB (5 parts) | 191.0 GB | Too big | Get file → | |
| IQ3_M | 189.6 GB (5 parts) | 191.1 GB | Too big | Get file → | |
| Q3_K_XL | 190.4 GB (5 parts) | 191.9 GB | Too big | Get file → | |
| IQ4_XS | 212.2 GB (6 parts) | 213.7 GB | Too big | Get file → | |
| IQ4_NL | 224.5 GB (6 parts) | 226.0 GB | Too big | Get file → | |
| Q4_0 | 225.4 GB (6 parts) | 226.9 GB | Too big | Get file → | |
| Q4_K_S | 232.8 GB (6 parts) | 234.3 GB | Too big | Get file → | |
| Q4_K_M | 241.8 GB (7 parts) | 243.3 GB | Too big | Get file → | |
| Q4_K_L | 242.6 GB (7 parts) | 244.1 GB | Too big | Get file → | |
| Q4_1 | 249.1 GB (7 parts) | 250.6 GB | Too big | Get file → | |
| Q5_K_S | 274.0 GB (7 parts) | 275.5 GB | Too big | Get file → | |
| Q5_K_M | 283.3 GB (8 parts) | 284.8 GB | Too big | Get file → | |
| Q6_K | 342.4 GB (9 parts) | 343.9 GB | Too big | Get file → | |
| Q8_0 | 421.6 GB (11 parts) | 423.1 GB | Too big | Get file → |
For 8‑GB VRAM, use the IQ1_S quant (file size 81.78 GB, memory 83.3 GB at 8k context). For 16‑GB VRAM, IQ2_XS (118.47 GB, 120 GB). For 24‑GB VRAM, IQ3_XS (181.44 GB, 183 GB). Larger VRAM allows higher‑quality quants such as Q4_K_M (241.81 GB, 243.3 GB).
What GPU do you need for Ornith 1.0 397B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 32 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
- 48 GB VRAM: the best quant that fits is the main file (1.1 GB file, about 2.6 GB in use).
Which Ornith 1.0 397B quant should you choose?
Q4_K_M is the usual starting point for Ornith 1.0 397B: it balances file size and quality well. Here it is a 241.8 GB file.
The Q8_0 file is 421.6 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is the first one in the table at 1.1 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Ornith 1.0 397B
How much VRAM does Ornith 1.0 397B need?
About 243.3 GB at Q4_K_M with an 8K context window. That is the 241.8 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Ornith 1.0 397B on an 8 GB GPU?
Yes, with the smaller files: the 1.1 GB file needs about 2.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Ornith 1.0 397B GGUF?
It depends on the quant: file is 1.1 GB, IQ1_S is 81.8 GB, IQ1_M is 91.4 GB, IQ2_XXS is 106.3 GB, IQ2_XS is 118.5 GB, IQ2_S is 120.5 GB and 21 more.
What context length does Ornith 1.0 397B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.0 GB at 8K and 4.0 GB at 32K.
Where can I download Ornith 1.0 397B GGUF?
The files are on Hugging Face in the bartowski/deepreinforce-ai_Ornith-1.0-397B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Ornith 1.0 397B use?
Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Ornith 1.0 397B on 8 GB VRAM?
Only with the IQ1_S quant, which requires about 83.3 GB of VRAM at 8k context. Standard full‑precision use needs far more.
Does the model support vision tasks?
The model has detected vision ability, but llama.cpp does not provide native vision support for this quantization.
What is the maximum context length?
The trained context length is 262144 tokens, but practical limits depend on VRAM and quant type.