calm3-22b-chat needs about 23.8 GB of memory at Q4_K_M (a 13.6 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 4 quants, from 13.6 GB (Q4_K_M) to 45.1 GB (BF16).
Calm3-22B-Chat GGUF model for Japanese and English
Calm3-22B-Chat is a 22.5B parameter bilingual model converted to GGUF format. It supports 16384 context tokens and is released under the apache-2.0 license. The GGUF files are optimized for llama.cpp usage.
Good at
- Japanese language tasks
- English language tasks
- Chat applications
Limits
- Pre‑tokenization may reduce performance
- No vision support
- Requires large VRAM for higher precision
Best for
- Japanese‑English dialogue systems
- Content generation in bilingual contexts
- Chatbot development
Who it’s for. Developers building bilingual chat applications or content generators who need a large language model in GGUF format. The model is suitable for users with access to llama.cpp and sufficient GPU memory.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 13.6 GB | 23.8 GB | Too big | Get file → | |
| Q5_K_M | 16.0 GB | 26.2 GB | Too big | Get file → | |
| Q6_K | 18.5 GB | 28.7 GB | Too big | Get file → | |
| BF16 | 45.1 GB | 55.3 GB | Too big | Get file → |
For 8 GB VRAM use Q4_K_M (13.64 GB file, ~23.8 GB VRAM). For 12 GB VRAM use Q5_K_M (16 GB file, ~26.2 GB VRAM). For 16 GB VRAM use Q6_K (18.5 GB file, ~28.7 GB VRAM). BF16 requires ~55.3 GB VRAM and is not recommended for typical setups.
What GPU do you need for calm3-22b-chat?
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: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (13.6 GB file, about 23.8 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (18.5 GB file, about 28.7 GB in use).
- 48 GB VRAM: the best quant that fits is Q6_K (18.5 GB file, about 28.7 GB in use).
Which calm3-22b-chat quant should you choose?
Q4_K_M is the usual starting point for calm3-22b-chat: it balances file size and quality well. Here it is a 13.6 GB file.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about calm3-22b-chat
How much VRAM does calm3-22b-chat need?
About 23.8 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 calm3-22b-chat on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 23.8 GB, so you would need a larger GPU or CPU offloading.
What is the file size of calm3-22b-chat GGUF?
It depends on the quant: Q4_K_M is 13.6 GB, Q5_K_M is 16.0 GB, Q6_K is 18.5 GB, BF16 is 45.1 GB.
What context length does calm3-22b-chat support?
The GGUF header lists a trained context length of 16,384 tokens (16K). The KV cache grows with the context you use: about 9.7 GB at 8K and 19.3 GB at 16K.
Where can I download calm3-22b-chat GGUF?
The files are on Hugging Face in the grapevine-AI/CALM3-22B-Chat-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 calm3-22b-chat 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 Calm3-22B-Chat on 8 GB VRAM?
Use the Q4_K_M quantized file; it requires about 23.8 GB VRAM, so 8 GB is insufficient.
Does the model support vision tasks?
No, the model has no vision capabilities.
What license applies to this model?
The model is released under the apache-2.0 license.