saihai v1 8b needs about 6.9 GB of memory at Q4_K_M (a 5.3 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 6 quants, from 3.5 GB (Q2_K) to 17.2 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.
Japanese decision model that returns yes/no, choice, and score probabilities
saihai v1 8b is a Japanese typed decision model based on llama. It answers questions in a single forward pass without generating text. The model is fine‑tuned from llm-jp-4.1-8b-thinking and released under Apache‑2.0.
Good at
- Yes/No probability (Noul)
- Choice selection
- Score rubric
Limits
- No vision support
- Requires 8k context for estimates
- No text generation
Best for
- Customer support decision making
- Refund eligibility checks
- Urgency scoring
Who it’s for. Developers building decision‑making tools that need fast, probabilistic answers in Japanese. The model is suitable for applications that require no text generation and can run on llama.cpp.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 3.5 GB | 5.1 GB | Fits | Get file → | |
| Q3_K_M | 4.4 GB | 5.9 GB | Fits | Get file → | |
| Q4_K_M | 5.3 GB | 6.9 GB | Fits | Get file → | |
| Q6_K | 7.1 GB | 8.6 GB | Too big | Get file → | |
| Q8_0 | 9.1 GB | 10.7 GB | Too big | Get file → | |
| BF16 | 17.2 GB | 18.8 GB | Too big | Get file → |
For the best quality, use Q8_0 (file size 9.13 GB, estimated 10.7 GB memory at 8k context). If memory is limited, Q4_K_M (5.3 GB, 6.9 GB memory) is recommended. Q2_K, Q3_K_M, and Q6_K are available for lower memory budgets but provide lower quality.
What GPU do you need for saihai v1 8b?
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 Q3_K_M (4.4 GB file, about 5.9 GB in use).
- 8 GB VRAM: the best quant that fits is Q4_K_M (5.3 GB file, about 6.9 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (9.1 GB file, about 10.7 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (9.1 GB file, about 10.7 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (17.2 GB file, about 18.8 GB in use).
Which saihai v1 8b quant should you choose?
Q4_K_M is the usual starting point for saihai v1 8b: it balances file size and quality well. Here it is a 5.3 GB file.
The Q8_0 file is 9.1 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 Q2_K at 3.5 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 saihai v1 8b
How much VRAM does saihai v1 8b need?
About 6.9 GB at Q4_K_M with an 8K context window. That is the 5.3 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 saihai v1 8b on an 8 GB GPU?
Yes, with Q4_K_M: the 5.3 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of saihai v1 8b GGUF?
It depends on the quant: Q2_K is 3.5 GB, Q3_K_M is 4.4 GB, Q4_K_M is 5.3 GB, Q6_K is 7.1 GB, Q8_0 is 9.1 GB, BF16 is 17.2 GB.
What context length does saihai v1 8b support?
The GGUF header lists a trained context length of 65,536 tokens (64K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.
Where can I download saihai v1 8b GGUF?
The files are on Hugging Face in the hotchpotch/saihai-v1-8b-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 saihai v1 8b 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 Q4_K_M on a machine with 8 GB VRAM?
Q4_K_M requires approximately 6.9 GB memory at 8k context, so it can run on a system with 8 GB VRAM, but VRAM usage may be higher due to overhead.
Does the model support English?
The model is trained on Japanese and English decision data, so it can handle both languages for decision tasks.
Is the model capable of generating text?
No, saihai v1 8b returns probabilities for Noul, Choice, and Score without generating free‑form text.