cerebras/Qwen3-Coder-REAP-25B-A3B needs about 16.5 GB of memory at Q4_K_M (a 15.2 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 28 quants, from 0.1 GB to 49.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.2 GB.
25B Qwen3-Coder compressed with REAP, 20% lighter, 262k token context
Qwen3-Coder-REAP-25B-A3B is a 25 billion‑parameter sparse Mixture‑of‑Experts model compressed from 30 B. It retains near‑lossless coding performance and supports a native context length of 262,144 tokens. The model is released under Apache 2.0 and is compatible with vLLM and llama.cpp.
Good at
- Code generation
- Agentic coding
- Function calling
Limits
- Large context length requires high VRAM
- No vision support
- Requires 8 k context for memory estimates
Best for
- Local code assistant deployments
- Research on sparse MoE models
- Resource‑constrained coding tasks
Who it’s for. Developers and researchers needing a compact, high‑capacity coding model that runs on local hardware. The model is suitable for environments where memory is limited but code generation quality must remain high.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.1 GB | 1.4 GB | Fits | Get file → | |
| IQ2_XXS | 6.2 GB | 7.5 GB | Tight | Get file → | |
| IQ2_XS | 7.1 GB | 8.4 GB | Too big | Get file → | |
| IQ2_S | 7.2 GB | 8.5 GB | Too big | Get file → | |
| IQ2_M | 8.1 GB | 9.4 GB | Too big | Get file → | |
| Q2_K | 8.9 GB | 10.3 GB | Too big | Get file → | |
| Q2_K_L | 9.2 GB | 10.6 GB | Too big | Get file → | |
| IQ3_XXS | 10.0 GB | 11.3 GB | Too big | Get file → | |
| IQ3_XS | 10.4 GB | 11.7 GB | Too big | Get file → | |
| Q3_K_S | 11.0 GB | 12.3 GB | Too big | Get file → | |
| IQ3_M | 11.5 GB | 12.8 GB | Too big | Get file → | |
| Q3_K_M | 11.5 GB | 12.8 GB | Too big | Get file → | |
| Q3_K_L | 11.9 GB | 13.2 GB | Too big | Get file → | |
| Q3_K_XL | 12.2 GB | 13.5 GB | Too big | Get file → | |
| IQ4_XS | 13.4 GB | 14.7 GB | Too big | Get file → | |
| IQ4_NL | 14.2 GB | 15.5 GB | Too big | Get file → | |
| Q4_0 | 14.4 GB | 15.7 GB | Too big | Get file → | |
| Q4_K_S | 14.7 GB | 16.0 GB | Too big | Get file → | |
| Q4_K_M | 15.2 GB | 16.5 GB | Too big | Get file → | |
| Q4_K_L | 15.4 GB | 16.7 GB | Too big | Get file → | |
| Q4_1 | 15.7 GB | 17.0 GB | Too big | Get file → | |
| Q5_K_S | 17.2 GB | 18.5 GB | Too big | Get file → | |
| Q5_K_M | 17.7 GB | 19.0 GB | Too big | Get file → | |
| Q5_K_L | 17.9 GB | 19.2 GB | Too big | Get file → | |
| Q6_K | 20.5 GB | 21.8 GB | Too big | Get file → | |
| Q6_K_L | 20.6 GB | 21.9 GB | Too big | Get file → | |
| Q8_0 | 26.5 GB | 27.8 GB | Too big | Get file → | |
| BF16 | 49.8 GB | 51.1 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (6.23 GB, ~7.5 GB memory at 8 k context). For 16 GB VRAM, IQ4_K_M (15.19 GB, ~16.5 GB memory) or Q4_K_L (15.42 GB, ~16.7 GB). For 24 GB VRAM, Q5_K_M (17.72 GB, ~19 GB memory) or Q6_K (20.45 GB, ~21.8 GB).
What GPU do you need for cerebras/Qwen3-Coder-REAP-25B-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: the best quant that fits is the main file (0.1 GB file, about 1.4 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 1.4 GB in use).
- 8 GB VRAM: the best quant that fits is IQ2_XXS (6.2 GB file, about 7.5 GB in use).
- 12 GB VRAM: the best quant that fits is IQ3_XS (10.4 GB file, about 11.7 GB in use).
- 16 GB VRAM: the best quant that fits is Q4_K_S (14.7 GB file, about 16.0 GB in use).
- 24 GB VRAM: the best quant that fits is Q6_K_L (20.6 GB file, about 21.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (26.5 GB file, about 27.8 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (26.5 GB file, about 27.8 GB in use).
Which cerebras/Qwen3-Coder-REAP-25B-A3B quant should you choose?
Q4_K_M is the usual starting point for cerebras/Qwen3-Coder-REAP-25B-A3B: it balances file size and quality well. Here it is a 15.2 GB file.
The Q8_0 file is 26.5 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 0.1 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 cerebras/Qwen3-Coder-REAP-25B-A3B
How much VRAM does cerebras/Qwen3-Coder-REAP-25B-A3B need?
About 16.5 GB at Q4_K_M with an 8K context window. That is the 15.2 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 cerebras/Qwen3-Coder-REAP-25B-A3B on an 8 GB GPU?
Yes, with IQ2_XXS: the 6.2 GB file needs about 7.5 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of cerebras/Qwen3-Coder-REAP-25B-A3B GGUF?
It depends on the quant: file is 0.1 GB, IQ2_XXS is 6.2 GB, IQ2_XS is 7.1 GB, IQ2_S is 7.2 GB, IQ2_M is 8.1 GB, Q2_K is 8.9 GB and 22 more.
What context length does cerebras/Qwen3-Coder-REAP-25B-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 0.8 GB at 8K and 3.2 GB at 32K.
Where can I download cerebras/Qwen3-Coder-REAP-25B-A3B GGUF?
The files are on Hugging Face in the bartowski/cerebras_Qwen3-Coder-REAP-25B-A3B-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 cerebras/Qwen3-Coder-REAP-25B-A3B 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 Qwen3-Coder-REAP-25B-A3B on 8 GB VRAM?
Yes, using the IQ2_XXS quantization which requires about 7.5 GB of VRAM at 8 k context.
Does the model support image inputs?
No, the model is a text‑only causal language model with no vision capabilities.
What context length does the model support?
The native context length is 262,144 tokens, extendable to 1 M tokens with YaRN.