KA

Kanana 2 1.3b Instruct

KananaGGUF by winwinplatformUpdated 2026-09-28
Download on Hugging Face
Parameters1.29B
Context32K
Layers32
Architecturekanana2tiny
Smallest quant0.0 GB
Quick answer

Kanana 2 1.3b Instruct needs about 2.4 GB of memory at Q4_K_M (a 0.9 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 15 quants, from 0.0 GB to 2.6 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Kanana 2 1.3b Instruct – compact instruction‑tuned model

Kanana 2 1.3b Instruct is a 1.29B parameter model built from Kanana‑2‑3B via cascade pruning and distillation. It supports up to 32K context tokens with sliding window attention. The model is available in multiple GGUF quantizations for on‑device inference.

Good at

  • Reasoning
  • Tool usage
  • Long‑context instruction following

Limits

  • No vision support
  • Requires patched llama.cpp for kanana2tiny
  • License is other

Best for

  • On‑device chat assistants
  • Lightweight reasoning tasks
  • Short‑form instruction following

Who it’s for. Developers building mobile or embedded LLM applications that need instruction following and reasoning within a 1.3B parameter budget. The model is suited for users who can compile patched llama.cpp and prefer a small, efficient GGUF format.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.0 GB 1.6 GB
Fits Get file →
Q3_K_M0.7 GB 2.3 GB
Fits Get file →
IQ4_XS0.8 GB 2.3 GB
Fits Get file →
Q4_00.8 GB 2.3 GB
Fits Get file →
Q4_00.8 GB 2.4 GB
Fits Get file →
IQ4_NL0.8 GB 2.4 GB
Fits Get file →
Q4_K_M0.9 GB 2.4 GB
Fits Get file →
Q4_K_M0.9 GB 2.4 GB
Fits Get file →
Q4_K_M0.9 GB 2.5 GB
Fits Get file →
Q5_K_M1.0 GB 2.5 GB
Fits Get file →
Q5_K_M1.0 GB 2.5 GB
Fits Get file →
Q6_K1.1 GB 2.7 GB
Fits Get file →
Q6_K1.1 GB 2.7 GB
Fits Get file →
Q8_01.4 GB 3.0 GB
Fits Get file →
F162.6 GB 4.2 GB
Fits Get file →

For 8‑GB VRAM, use Q5_K_M (0.97 GB, 2.5 GB memory at 8K context) or Q4_K_M (0.85 GB, 2.4 GB memory). For 4‑GB VRAM, Q3_K_M (0.71 GB, 2.3 GB memory) is recommended. Larger VRAM allows Q8_0 (1.38 GB, 3 GB memory) or F16 (2.59 GB, 4.2 GB memory).

What GPU do you need for Kanana 2 1.3b Instruct?

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 Q8_0 (1.4 GB file, about 3.0 GB in use).
  • 6 GB VRAM: the best quant that fits is F16 (2.6 GB file, about 4.2 GB in use).

Which Kanana 2 1.3b Instruct quant should you choose?

Q4_K_M is the usual starting point for Kanana 2 1.3b Instruct: it balances file size and quality well. Here it is a 0.9 GB file.

The Q8_0 file is 1.4 GB, about 1.6 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 0.0 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

Architecturekanana2tiny
Parameters1.29B
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized bywinwinplatform
FamilyKanana
Licenseother
Downloads (HF)255
Last updated2026-09-28

Frequently asked questions about Kanana 2 1.3b Instruct

How much VRAM does Kanana 2 1.3b Instruct need?

About 2.4 GB at Q4_K_M with an 8K context window. That is the 0.9 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 Kanana 2 1.3b Instruct on an 8 GB GPU?

Yes, with F16: the 2.6 GB file needs about 4.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Kanana 2 1.3b Instruct GGUF?

It depends on the quant: file is 0.0 GB, Q3_K_M is 0.7 GB, IQ4_XS is 0.8 GB, Q4_0 is 0.8 GB, Q4_0 is 0.8 GB, IQ4_NL is 0.8 GB and 9 more.

What context length does Kanana 2 1.3b Instruct support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download Kanana 2 1.3b Instruct GGUF?

The files are on Hugging Face in the winwinplatform/Kanana-2-1.3b-instruct-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 Kanana 2 1.3b Instruct use?

Hugging Face lists the license as other. 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 Kanana 2 1.3b Instruct on 8 GB VRAM?

Yes, using Q5_K_M or Q4_K_M quantizations will fit within 8 GB VRAM, with estimated 2.5 GB or 2.4 GB memory usage at 8K context.

Does this model support vision or multimodal input?

No, the model only handles text and has no vision capabilities.

What version of llama.cpp is required?

A patched version that supports the kanana2tiny architecture is required; the standard llama.cpp will return an unknown model architecture error.

Other GGUFs of kanana-2-1.3b-instruct

Similar models

Browse all →

More Kanana models

All Kanana →

Other tiny models

All tiny models →