Ornith-1.5-35B-A3B needs about 22.2 GB of memory at Q4_K_M (a 21.0 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 3 quants, from 21.0 GB (Q4_K_M) to 30.9 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 2.8 GB.
Ornith-1.5-35B-A3B – a vision‑capable, 35B‑parameter mixture‑of‑experts model
Ornith-1.5-35B-A3B is a 35.5‑billion‑parameter model that activates only ~3B parameters per token. It is built on the qwen35moe architecture and supports vision tasks. The model is released under the MIT license.
Good at
- Vision‑enabled language tasks
- Mixture‑of‑experts efficiency
- Large context handling
Limits
- No text‑only inference on very small VRAM
- Requires at least 8 k context for memory estimates
- No explicit tool integration listed
Best for
- Vision‑enabled chat applications
- Code generation with expert routing
- Long‑form document analysis
Who it’s for. Developers building vision‑enabled LLMs who need a large‑parameter model with efficient expert routing. The MIT license allows free use in research and commercial projects.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 21.0 GB | 22.2 GB | Too big | Get file → | |
| Q5_K_M | 24.6 GB | 25.8 GB | Too big | Get file → | |
| Q6_K | 30.9 GB | 32.1 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (21.02 GB file, ~22.2 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (24.56 GB file, ~25.8 GB memory). For 16 GB VRAM, Q6_K (30.94 GB file, ~32.1 GB memory). Memory estimates are approximate.
What GPU do you need for Ornith-1.5-35B-A3B?
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: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (21.0 GB file, about 22.2 GB in use).
- 32 GB VRAM: the best quant that fits is Q5_K_M (24.6 GB file, about 25.8 GB in use).
- 48 GB VRAM: the best quant that fits is Q6_K (30.9 GB file, about 32.1 GB in use).
Which Ornith-1.5-35B-A3B quant should you choose?
Q4_K_M is the usual starting point for Ornith-1.5-35B-A3B: it balances file size and quality well. Here it is a 21.0 GB file.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Ornith-1.5-35B-A3B
How much VRAM does Ornith-1.5-35B-A3B need?
About 22.2 GB at Q4_K_M with an 8K context window. That is the 21.0 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-1.5-35B-A3B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 22.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Ornith-1.5-35B-A3B GGUF?
It depends on the quant: Q4_K_M is 21.0 GB, Q5_K_M is 24.6 GB, Q6_K is 30.9 GB.
What context length does Ornith-1.5-35B-A3B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.8 GB at 32K.
Where can I download Ornith-1.5-35B-A3B GGUF?
The files are on Hugging Face in the lmcoleman/Ornith-1.5-35B-A3B-MagicQuant-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-1.5-35B-A3B 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-1.5-35B-A3B on 8 GB VRAM?
Yes, using the Q4_K_M quantization, which requires about 22.2 GB of VRAM at 8 k context.
Does this model support vision tasks?
Yes, the model has detected vision capabilities and is vision‑capable.