RecursiveSI flint needs about 26.7 GB of memory at Q4_K_M (a 25.3 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 26 quants, from 0.5 GB to 79.6 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 3.3 GB.
39.8B‑parameter mixture‑of‑experts coding model
RecursiveSI flint is a 39.8 billion‑parameter mixture‑of‑experts language model designed for coding tasks. It supports text input and output, reasoning, and multi‑turn tool use. The model is released under the apache‑2.0 license.
Good at
- Code investigation and bug fixing
- Feature implementation and test generation
- Multi‑turn tool‑based reasoning
Limits
- Requires large VRAM for full precision
- No vision support
- Active parameters per token are ~2.5 billion
Best for
- Developers debugging complex codebases
- Automated test writing
- Agentic coding assistants
Who it’s for. Engineers and researchers who need a powerful coding assistant that can reason across large contexts and use tools. The model is suitable for environments that can provide the necessary GPU memory and support for llama.cpp or SGLang.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.5 GB | 1.8 GB | Fits | Get file → | |
| IQ2_XXS | 12.0 GB | 13.4 GB | Too big | Get file → | |
| IQ2_XS | 12.7 GB | 14.1 GB | Too big | Get file → | |
| IQ2_S | 13.4 GB | 14.7 GB | Too big | Get file → | |
| IQ2_M | 14.9 GB | 16.2 GB | Too big | Get file → | |
| Q2_K | 15.5 GB | 16.8 GB | Too big | Get file → | |
| IQ3_XXS | 17.5 GB | 18.8 GB | Too big | Get file → | |
| IQ3_XS | 18.3 GB | 19.6 GB | Too big | Get file → | |
| Q3_K_S | 18.4 GB | 19.8 GB | Too big | Get file → | |
| Q3_K_M | 19.5 GB | 20.9 GB | Too big | Get file → | |
| Q3_K_L | 20.6 GB | 21.9 GB | Too big | Get file → | |
| IQ3_M | 21.1 GB | 22.4 GB | Too big | Get file → | |
| IQ4_XS | 22.4 GB | 23.8 GB | Too big | Get file → | |
| Q4_0 | 22.8 GB | 24.2 GB | Too big | Get file → | |
| Q4_K_S | 23.8 GB | 25.2 GB | Too big | Get file → | |
| Q4_1 | 25.2 GB | 26.5 GB | Too big | Get file → | |
| IQ4_NL | 25.3 GB | 26.6 GB | Too big | Get file → | |
| Q4_K_M | 25.3 GB | 26.7 GB | Too big | Get file → | |
| Q4_K_L | 27.3 GB | 28.6 GB | Too big | Get file → | |
| Q5_K_S | 28.6 GB | 29.9 GB | Too big | Get file → | |
| Q5_K_M | 30.8 GB | 32.1 GB | Too big | Get file → | |
| Q6_K_S | 33.6 GB | 34.9 GB | Too big | Get file → | |
| Q6_K | 35.2 GB | 36.5 GB | Too big | Get file → | |
| Q6_K_L | 37.0 GB | 38.3 GB | Too big | Get file → | |
| Q8_0 | 42.4 GB | 43.7 GB | Too big | Get file → | |
| BF16 | 79.6 GB (2 parts) | 80.9 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (12.03 GB) or IQ2_XS (12.74 GB). For 16 GB VRAM, IQ4_K_M (25.35 GB) offers a good balance of size and performance. For 32 GB VRAM, Q6_K (35.2 GB) or Q6_K_L (36.95 GB) provide higher fidelity.
What GPU do you need for RecursiveSI flint?
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 (0.5 GB file, about 1.8 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
- 12 GB VRAM: the best quant that fits is the main file (0.5 GB file, about 1.8 GB in use).
- 16 GB VRAM: the best quant that fits is IQ2_S (13.4 GB file, about 14.7 GB in use).
- 24 GB VRAM: the best quant that fits is IQ4_XS (22.4 GB file, about 23.8 GB in use).
- 32 GB VRAM: the best quant that fits is Q5_K_S (28.6 GB file, about 29.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (42.4 GB file, about 43.7 GB in use).
Which RecursiveSI flint quant should you choose?
Q4_K_M is the usual starting point for RecursiveSI flint: it balances file size and quality well. Here it is a 25.3 GB file.
The Q8_0 file is 42.4 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 0.5 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 RecursiveSI flint
How much VRAM does RecursiveSI flint need?
About 26.7 GB at Q4_K_M with an 8K context window. That is the 25.3 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 RecursiveSI flint on an 8 GB GPU?
Yes, with the smaller files: the 0.5 GB file needs about 1.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of RecursiveSI flint GGUF?
It depends on the quant: file is 0.5 GB, IQ2_XXS is 12.0 GB, IQ2_XS is 12.7 GB, IQ2_S is 13.4 GB, IQ2_M is 14.9 GB, Q2_K is 15.5 GB and 20 more.
What context length does RecursiveSI flint 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 3.3 GB at 32K.
Where can I download RecursiveSI flint GGUF?
The files are on Hugging Face in the bartowski/RecursiveSI_flint-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 RecursiveSI flint 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 flint on 8 GB VRAM?
Yes, use the IQ2_XXS or IQ2_XS quantized files, which are 12.03 GB and 12.74 GB respectively.
Does flint support vision tasks?
No, flint only supports text input and output.
What is the context length of flint?
Flint was trained with a context length of 262144 tokens.