Triangulum 10B needs about 8.1 GB of memory at Q4_K_M (a 6.3 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 24 quants, from 3.3 GB (IQ2_S) to 20.6 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.
Triangulum 10B – 10.3B‑parameter LLaMA‑based model for reasoning
Triangulum 10B is a 10.3B‑parameter LLaMA‑based model trained on synthetic chain‑of‑thought data. It is instruction‑tuned with SFT and RLHF for helpfulness and safety. The model supports a 32768‑token context window.
Good at
- Reasoning tasks
- Multilingual content generation
- Dialogue systems
Limits
- No vision support
- Requires 8‑k context for memory estimates
- License: llama3.1
Best for
- Complex reasoning queries
- Multilingual Q&A
- Text summarization
Who it’s for. Developers building multilingual LLM applications that need strong reasoning capabilities. Researchers exploring instruction‑tuned LLaMA variants.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_S | 3.3 GB | 5.2 GB | Fits | Get file → | |
| IQ2_M | 3.6 GB | 5.4 GB | Fits | Get file → | |
| Q2_K | 3.9 GB | 5.8 GB | Fits | Get file → | |
| Q2_K_L | 4.3 GB | 6.2 GB | Fits | Get file → | |
| IQ3_XS | 4.4 GB | 6.2 GB | Fits | Get file → | |
| Q3_K_S | 4.6 GB | 6.4 GB | Fits | Get file → | |
| IQ3_M | 4.7 GB | 6.5 GB | Fits | Get file → | |
| Q3_K_M | 5.1 GB | 6.9 GB | Fits | Get file → | |
| Q3_K_L | 5.5 GB | 7.3 GB | Tight | Get file → | |
| IQ4_XS | 5.6 GB | 7.4 GB | Tight | Get file → | |
| Q3_K_XL | 5.8 GB | 7.6 GB | Tight | Get file → | |
| IQ4_NL | 5.9 GB | 7.7 GB | Tight | Get file → | |
| Q4_0 | 5.9 GB | 7.8 GB | Tight | Get file → | |
| Q4_K_S | 6.0 GB | 7.8 GB | Tight | Get file → | |
| Q4_K_M | 6.3 GB | 8.1 GB | Too big | Get file → | |
| Q4_1 | 6.5 GB | 8.4 GB | Too big | Get file → | |
| Q4_K_L | 6.6 GB | 8.4 GB | Too big | Get file → | |
| Q5_K_S | 7.1 GB | 9.0 GB | Too big | Get file → | |
| Q5_K_M | 7.3 GB | 9.2 GB | Too big | Get file → | |
| Q5_K_L | 7.6 GB | 9.4 GB | Too big | Get file → | |
| Q6_K | 8.5 GB | 10.3 GB | Too big | Get file → | |
| Q6_K_L | 8.7 GB | 10.5 GB | Too big | Get file → | |
| Q8_0 | 11.0 GB | 12.8 GB | Too big | Get file → | |
| F16 | 20.6 GB | 22.5 GB | Too big | Get file → |
For 8‑GB VRAM, use IQ2_S (3.32 GB, 5.2 GB memory) or IQ2_M (3.59 GB, 5.4 GB memory). For 16‑GB VRAM, Q4_K_M (6.29 GB, 8.1 GB memory) or Q5_K_S (7.14 GB, 9 GB memory) provide a good balance of quality and size. For 24‑GB VRAM, Q6_K (8.46 GB, 10.3 GB memory) or Q8_0 (10.96 GB, 12.8 GB memory) are suitable.
What GPU do you need for Triangulum 10B?
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: the best quant that fits is Q2_K (3.9 GB file, about 5.8 GB in use).
- 8 GB VRAM: the best quant that fits is Q4_K_S (6.0 GB file, about 7.8 GB in use).
- 12 GB VRAM: the best quant that fits is Q6_K_L (8.7 GB file, about 10.5 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (11.0 GB file, about 12.8 GB in use).
- 24 GB VRAM: the best quant that fits is F16 (20.6 GB file, about 22.5 GB in use).
Which Triangulum 10B quant should you choose?
Q4_K_M is the usual starting point for Triangulum 10B: it balances file size and quality well. Here it is a 6.3 GB file.
The Q8_0 file is 11.0 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 IQ2_S at 3.3 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 Triangulum 10B
How much VRAM does Triangulum 10B need?
About 8.1 GB at Q4_K_M with an 8K context window. That is the 6.3 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 Triangulum 10B on an 8 GB GPU?
Yes, with Q4_K_S: the 6.0 GB file needs about 7.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Triangulum 10B GGUF?
It depends on the quant: IQ2_S is 3.3 GB, IQ2_M is 3.6 GB, Q2_K is 3.9 GB, Q2_K_L is 4.3 GB, IQ3_XS is 4.4 GB, Q3_K_S is 4.6 GB and 18 more.
What context length does Triangulum 10B support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.3 GB at 8K and 5.4 GB at 32K.
Where can I download Triangulum 10B GGUF?
The files are on Hugging Face in the bartowski/Triangulum-10B-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 Triangulum 10B use?
Hugging Face lists the license as llama3.1. 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 Triangulum 10B on 8 GB VRAM?
Yes, use IQ2_S or IQ2_M quantizations, which are 3.32 GB and 3.59 GB respectively and require about 5.2–5.4 GB of VRAM at 8‑k context.
Does Triangulum 10B support vision?
No, the model has no vision capabilities.