Reka Flash 3 needs about 15.2 GB of memory at Q4_K_M (a 13.6 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 28 quants, from 6.6 GB (IQ1_S) to 41.8 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.4 GB.
Reka Flash 3 – 20.9B parameter reasoning model in GGUF format
Reka Flash 3 is a 20.9B parameter general‑purpose reasoning model released in GGUF format. It was trained from scratch on synthetic and public datasets, followed by RLOO with model‑based and rule‑based rewards. The model supports a 32768 token context window and is licensed under apache‑2.0.
Good at
- Reasoning tasks
- Low‑latency inference
- On‑device deployment
Limits
- No vision support
- Requires large VRAM for higher quantizations
- No explicit language support list
Best for
- Chatbots needing reasoning
- Code generation with reasoning traces
- Research assistants
Who it’s for. Developers building reasoning‑heavy applications who need a large open model that can run locally. The model is suitable for teams that require a 32768 token context and can manage the VRAM demands of the chosen quantization.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 6.6 GB | 8.2 GB | Too big | Get file → | |
| IQ1_M | 6.9 GB | 8.5 GB | Too big | Get file → | |
| IQ2_XXS | 7.4 GB | 9.0 GB | Too big | Get file → | |
| IQ2_XS | 7.8 GB | 9.4 GB | Too big | Get file → | |
| IQ2_S | 8.1 GB | 9.7 GB | Too big | Get file → | |
| IQ2_M | 8.5 GB | 10.1 GB | Too big | Get file → | |
| Q2_K_S | 8.5 GB | 10.1 GB | Too big | Get file → | |
| Q2_K | 8.6 GB | 10.2 GB | Too big | Get file → | |
| IQ3_XXS | 9.2 GB | 10.8 GB | Too big | Get file → | |
| IQ3_XS | 9.5 GB | 11.1 GB | Too big | Get file → | |
| Q3_K_S | 9.9 GB | 11.5 GB | Too big | Get file → | |
| IQ3_S | 10.0 GB | 11.6 GB | Too big | Get file → | |
| IQ3_M | 10.3 GB | 11.9 GB | Too big | Get file → | |
| Q3_K_M | 10.9 GB | 12.5 GB | Too big | Get file → | |
| Q3_K_L | 11.4 GB | 13.0 GB | Too big | Get file → | |
| IQ4_XS | 11.5 GB | 13.1 GB | Too big | Get file → | |
| Q4_0 | 11.9 GB | 13.5 GB | Too big | Get file → | |
| IQ4_NL | 11.9 GB | 13.6 GB | Too big | Get file → | |
| Q4_K_S | 12.6 GB | 14.2 GB | Too big | Get file → | |
| Q4_1 | 13.2 GB | 14.8 GB | Too big | Get file → | |
| Q4_K_M | 13.6 GB | 15.2 GB | Too big | Get file → | |
| Q5_0 | 14.5 GB | 16.1 GB | Too big | Get file → | |
| Q5_K_S | 14.8 GB | 16.4 GB | Too big | Get file → | |
| Q5_K_M | 15.6 GB | 17.2 GB | Too big | Get file → | |
| Q5_1 | 15.7 GB | 17.3 GB | Too big | Get file → | |
| Q6_K | 18.4 GB | 20.0 GB | Too big | Get file → | |
| Q8_0 | 22.2 GB | 23.8 GB | Too big | Get file → | |
| F16 | 41.8 GB | 43.4 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (6.6 GB file, 8.2 GB memory at 8 k context). For 12 GB VRAM, use Q2_K (8.63 GB file, 10.2 GB memory). For 16 GB VRAM, use Q3_K_M (10.86 GB file, 12.5 GB memory). Estimates are approximate.
What GPU do you need for Reka Flash 3?
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: the best quant that fits is IQ3_M (10.3 GB file, about 11.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q4_K_M (13.6 GB file, about 15.2 GB in use).
- 24 GB VRAM: the best quant that fits is Q8_0 (22.2 GB file, about 23.8 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (22.2 GB file, about 23.8 GB in use).
- 48 GB VRAM: the best quant that fits is F16 (41.8 GB file, about 43.4 GB in use).
Which Reka Flash 3 quant should you choose?
Q4_K_M is the usual starting point for Reka Flash 3: it balances file size and quality well. Here it is a 13.6 GB file.
The Q8_0 file is 22.2 GB, about 1.6 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 IQ1_S at 6.6 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Reka Flash 3
How much VRAM does Reka Flash 3 need?
About 15.2 GB at Q4_K_M with an 8K context window. That is the 13.6 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 Reka Flash 3 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 8.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Reka Flash 3 GGUF?
It depends on the quant: IQ1_S is 6.6 GB, IQ1_M is 6.9 GB, IQ2_XXS is 7.4 GB, IQ2_XS is 7.8 GB, IQ2_S is 8.1 GB, IQ2_M is 8.5 GB and 22 more.
What context length does Reka Flash 3 support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.4 GB at 32K.
Where can I download Reka Flash 3 GGUF?
The files are on Hugging Face in the brittlewis12/reka-flash-3-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 Reka Flash 3 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 Reka Flash 3 on 8 GB VRAM?
Yes, the IQ1_S quantization requires 8.2 GB memory at 8 k context, which fits within 8 GB VRAM with some overhead.
Does the model support image input?
No, the model has no vision capabilities and only processes text.
What license applies to the model?
The model is released under the apache‑2.0 license.