NuminaMath-7B-TIR needs about 7.8 GB of memory at IQ3_M (a 3.3 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 25 quants, from 1.7 GB (IQ1_S) to 13.8 GB.
NuminaMath‑7B‑TIR: a 7B‑parameter math LLM fine‑tuned for tool‑integrated reasoning
NuminaMath‑7B‑TIR is a 6.91‑billion‑parameter language model fine‑tuned in two stages to solve math problems. It uses chain‑of‑thought reasoning and Python REPL execution to compute intermediate results. The model is released under Apache 2.0.
Good at
- Tool‑integrated math reasoning
- Chain‑of‑thought problem solving
- Python code generation
Limits
- No vision support
- Requires 8 k context for memory estimates
- License is Apache 2.0
Best for
- Math competition problem solving
- Educational math tutoring
- Research on tool‑integrated reasoning
Who it’s for. Developers and researchers needing a math‑focused LLM that can execute Python code for intermediate calculations. The model is suitable for educational tools and competition preparation.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 1.7 GB | 4.2 GB | Fits | Get file → | |
| IQ1_M | 1.8 GB | 4.4 GB | Fits | Get file → | |
| IQ2_XXS | 2.0 GB | 4.6 GB | Fits | Get file → | |
| IQ2_XS | 2.2 GB | 4.7 GB | Fits | Get file → | |
| IQ2_S | 2.4 GB | 4.9 GB | Fits | Get file → | |
| Q2_K_S | 2.5 GB | 5.0 GB | Fits | Get file → | |
| IQ2_M | 2.5 GB | 5.1 GB | Fits | Get file → | |
| Q2_K | 2.7 GB | 5.2 GB | Fits | Get file → | |
| IQ3_XXS | 2.8 GB | 5.3 GB | Fits | Get file → | |
| IQ3_XS | 3.0 GB | 5.5 GB | Fits | Get file → | |
| IQ3_S | 3.1 GB | 5.7 GB | Fits | Get file → | |
| Q3_K_S | 3.1 GB | 5.7 GB | Fits | Get file → | |
| IQ3_M | 3.3 GB | 5.8 GB | Fits | Get file → | |
| Q3_K | 3.5 GB | 6.0 GB | Fits | Get file → | |
| Q3_K_L | 3.7 GB | 6.3 GB | Fits | Get file → | |
| IQ4_XS | 3.8 GB | 6.3 GB | Fits | Get file → | |
| IQ4_NL | 4.0 GB | 6.5 GB | Fits | Get file → | |
| Q4_K_S | 4.0 GB | 6.5 GB | Fits | Get file → | |
| Q4_K | 4.2 GB | 6.7 GB | Fits | Get file → | |
| Q5_K_S | 4.8 GB | 7.3 GB | Tight | Get file → | |
| Q5_K | 4.9 GB | 7.4 GB | Tight | Get file → | |
| Q6_K | 5.7 GB | 8.2 GB | Too big | Get file → | |
| Q8_0 | 7.3 GB | 9.9 GB | Too big | Get file → | |
| BF16 | 13.8 GB | 16.3 GB | Too big | Get file → | |
| — | 13.8 GB | 16.3 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (1.73 GB) or IQ1_M (1.85 GB). For 16 GB VRAM, Q4_K (4.22 GB) or Q5_K (4.93 GB) provide a good balance of size and performance. For larger budgets, Q6_K (5.67 GB) or Q8_0 (7.35 GB) offer higher fidelity.
What GPU do you need for NuminaMath-7B-TIR?
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: the best quant that fits is Q3_K (3.5 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (7.3 GB file, about 11.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (7.3 GB file, about 11.9 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (13.8 GB file, about 18.4 GB in use).
Which NuminaMath-7B-TIR quant should you choose?
IQ3_M (3.3 GB) is the middle option in this repository. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
If memory is tight, the smallest file is IQ1_S at 1.7 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 NuminaMath-7B-TIR
How much VRAM does NuminaMath-7B-TIR need?
About 7.8 GB at IQ3_M with an 8K context window. That is the 3.3 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 NuminaMath-7B-TIR on an 8 GB GPU?
Yes, with Q3_K: the 3.5 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of NuminaMath-7B-TIR GGUF?
It depends on the quant: IQ1_S is 1.7 GB, IQ1_M is 1.8 GB, IQ2_XXS is 2.0 GB, IQ2_XS is 2.2 GB, IQ2_S is 2.4 GB, Q2_K_S is 2.5 GB and 19 more.
What context length does NuminaMath-7B-TIR support?
The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 4.0 GB at 8K and 2.0 GB at 4K.
Where can I download NuminaMath-7B-TIR GGUF?
The files are on Hugging Face in the legraphista/NuminaMath-7B-TIR-IMat-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 NuminaMath-7B-TIR 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 NuminaMath‑7B‑TIR on 8 GB VRAM?
Yes, use a quantization such as IQ1_S or IQ1_M, which are under 2 GB file size and require about 6.3–6.4 GB memory at 8 k context.
Does the model support vision inputs?
No, the model has no vision capabilities.
What license applies to this model?
The model is released under the Apache 2.0 license.