AREX Base needs about 76.3 GB of memory at Q4_K_M (a 74.9 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 29 quants, from 0.4 GB to 244.3 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.
AREX Base – a 122B‑parameter Mixture‑of‑Experts model for long‑horizon research
AREX Base is built on Qwen3.5‑122B‑A10B and uses a Mixture‑of‑Experts architecture. It supports a context length of 262,144 tokens and is designed for recursive self‑improvement and verification‑guided research.
Good at
- Long‑horizon reasoning
- Multi‑round tool use
- Evidence‑grounded answers
Limits
- No vision support
- Requires large VRAM for high‑quality quants
- Apache‑2.0 license limits commercial use
Best for
- Research agents
- Complex query answering
- Tool‑integrated workflows
Who it’s for. Developers building autonomous research agents or advanced question‑answering systems. The model is suitable for applications that need deep reasoning and iterative evidence gathering.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.4 GB | 1.7 GB | Fits | Get file → | |
| IQ1_S | 26.2 GB | 27.5 GB | Too big | Get file → | |
| IQ1_M | 29.1 GB | 30.4 GB | Too big | Get file → | |
| IQ2_XXS | 33.6 GB | 34.9 GB | Too big | Get file → | |
| IQ2_XS | 37.3 GB | 38.6 GB | Too big | Get file → | |
| IQ2_S | 38.0 GB | 39.3 GB | Too big | Get file → | |
| IQ2_M | 41.8 GB | 43.1 GB | Too big | Get file → | |
| Q2_K | 43.7 GB | 45.0 GB | Too big | Get file → | |
| Q2_K_L | 44.5 GB | 45.8 GB | Too big | Get file → | |
| IQ3_XXS | 51.8 GB | 53.1 GB | Too big | Get file → | |
| Q3_K_S | 54.0 GB (2 parts) | 55.3 GB | Too big | Get file → | |
| IQ3_XS | 56.5 GB (2 parts) | 57.8 GB | Too big | Get file → | |
| Q3_K_M | 56.6 GB (2 parts) | 57.9 GB | Too big | Get file → | |
| Q3_K_L | 58.9 GB (2 parts) | 60.2 GB | Too big | Get file → | |
| IQ3_M | 59.0 GB (2 parts) | 60.3 GB | Too big | Get file → | |
| Q3_K_XL | 59.6 GB (2 parts) | 60.9 GB | Too big | Get file → | |
| IQ4_XS | 65.8 GB (2 parts) | 67.1 GB | Too big | Get file → | |
| IQ4_NL | 69.6 GB (2 parts) | 70.9 GB | Too big | Get file → | |
| Q4_0 | 69.8 GB (2 parts) | 71.2 GB | Too big | Get file → | |
| Q4_K_S | 72.1 GB (2 parts) | 73.4 GB | Too big | Get file → | |
| Q4_K_M | 74.9 GB (2 parts) | 76.3 GB | Too big | Get file → | |
| Q4_K_L | 75.5 GB (2 parts) | 76.8 GB | Too big | Get file → | |
| Q4_1 | 77.1 GB (2 parts) | 78.4 GB | Too big | Get file → | |
| Q5_K_S | 84.8 GB (3 parts) | 86.1 GB | Too big | Get file → | |
| Q5_K_M | 87.8 GB (3 parts) | 89.1 GB | Too big | Get file → | |
| Q5_K_L | 88.2 GB (3 parts) | 89.5 GB | Too big | Get file → | |
| Q6_K | 105.7 GB (3 parts) | 107.0 GB | Too big | Get file → | |
| Q8_0 | 129.9 GB (4 parts) | 131.2 GB | Too big | Get file → | |
| BF16 | 244.3 GB (7 parts) | 245.6 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (26.22 GB file, 27.5 GB memory at 8k context). For 16 GB VRAM, IQ2_XXS (33.62 GB file, 34.9 GB memory) or Q2_K (43.73 GB file, 45 GB memory). For 24 GB VRAM, Q3_K_S (53.99 GB file, 55.3 GB memory) or Q3_K_M (56.56 GB file, 57.9 GB memory).
What GPU do you need for AREX Base?
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 (0.4 GB file, about 1.7 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.7 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.7 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.7 GB in use).
- 16 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.7 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.7 GB in use).
- 32 GB VRAM: the best quant that fits is IQ1_M (29.1 GB file, about 30.4 GB in use).
- 48 GB VRAM: the best quant that fits is Q2_K_L (44.5 GB file, about 45.8 GB in use).
Which AREX Base quant should you choose?
Q4_K_M is the usual starting point for AREX Base: it balances file size and quality well. Here it is a 74.9 GB file.
The Q8_0 file is 129.9 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 0.4 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 AREX Base
How much VRAM does AREX Base need?
About 76.3 GB at Q4_K_M with an 8K context window. That is the 74.9 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 AREX Base on an 8 GB GPU?
Yes, with the smaller files: the 0.4 GB file needs about 1.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of AREX Base GGUF?
It depends on the quant: file is 0.4 GB, IQ1_S is 26.2 GB, IQ1_M is 29.1 GB, IQ2_XXS is 33.6 GB, IQ2_XS is 37.3 GB, IQ2_S is 38.0 GB and 23 more.
What context length does AREX Base support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download AREX Base GGUF?
The files are on Hugging Face in the bartowski/BAAI_AREX-Base-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 AREX Base use?
Hugging Face lists the license as apache-2.0. 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 AREX Base on 8 GB VRAM?
Only the lowest‑quality quants such as IQ1_S or IQ1_M can fit within 8 GB VRAM, but performance may be limited.
Does AREX Base support vision?
No, the model does not have vision capabilities.