GR

granite-4.0-tiny-preview

GraniteGGUF by MungertUpdated 2025-09-24
Download on Hugging Face
Parameters6.67B
Context1024K
Layers40
Architecturegranitehybrid
Smallest quant1.5 GB
Quick answer

granite-4.0-tiny-preview needs about 6.7 GB of memory at Q4_K_M (a 4.2 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 27 quants, from 1.5 GB (TQ1_0) to 13.4 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.1 GB.

Granite-4.0-Tiny-Preview: 6.67B hybrid MoE LLM with 128k context

Granite-4.0-Tiny-Preview is a 6.67B parameter hybrid mixture-of-experts language model. It supports reasoning and tool use and offers a 1,048,576 token context window. The model is released under Apache‑2.0 and is available as GGUF files.

Good at

  • Reasoning
  • Tool use
  • Long‑context handling

Limits

  • No vision support
  • Requires large VRAM for higher quant levels
  • License is Apache‑2.0

Best for

  • Summarization
  • Question answering
  • Text classification

Who it’s for. Developers and researchers needing a compact yet powerful LLM for text‑to‑text tasks. The model is suitable for fine‑tuning on specialized domains.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
TQ1_01.5 GB 4.0 GB
Fits Get file →
TQ2_01.8 GB 4.3 GB
Fits Get file →
IQ2_XS2.1 GB 4.6 GB
Fits Get file →
IQ2_S2.2 GB 4.7 GB
Fits Get file →
IQ2_M2.3 GB 4.9 GB
Fits Get file →
Q2_K_M2.5 GB 5.0 GB
Fits Get file →
IQ3_XXS2.8 GB 5.3 GB
Fits Get file →
IQ3_XS2.8 GB 5.3 GB
Fits Get file →
IQ3_S3.1 GB 5.6 GB
Fits Get file →
IQ3_M3.1 GB 5.6 GB
Fits Get file →
Q3_K_S3.3 GB 5.8 GB
Fits Get file →
Q3_K_M3.3 GB 5.8 GB
Fits Get file →
IQ4_XS3.6 GB 6.1 GB
Fits Get file →
Q4_03.8 GB 6.3 GB
Fits Get file →
IQ4_NL3.8 GB 6.3 GB
Fits Get file →
Q4_K_S3.9 GB 6.5 GB
Fits Get file →
Q4_K_M4.2 GB 6.7 GB
Fits Get file →
Q4_14.2 GB 6.7 GB
Fits Get file →
Q5_04.6 GB 7.1 GB
Fits Get file →
Q5_K_S4.8 GB 7.3 GB
Tight Get file →
Q5_K_M4.9 GB 7.4 GB
Tight Get file →
Q5_15.0 GB 7.5 GB
Tight Get file →
Q6_K_M5.5 GB 8.0 GB
Too big Get file →
Q8_07.1 GB 9.6 GB
Too big Get file →
BF167.2 GB 9.7 GB
Too big Get file →
F167.2 GB 9.7 GB
Too big Get file →
BF1613.4 GB 15.9 GB
Too big Get file →

For 8 GB VRAM, use TQ1_0 (1.47 GB, 4 GB memory at 8k context) or TQ2_0 (1.78 GB, 4.3 GB memory). If 12 GB VRAM is available, Q4_K_M (4.15 GB, 6.7 GB memory) offers better precision. Higher quants like Q8_0 need 9.6 GB memory and are suitable for 16 GB VRAM setups.

What GPU do you need for granite-4.0-tiny-preview?

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 TQ1_0 (1.5 GB file, about 4.0 GB in use).
  • 6 GB VRAM: the best quant that fits is Q3_K_M (3.3 GB file, about 5.8 GB in use).
  • 8 GB VRAM: the best quant that fits is Q6_K_M (5.5 GB file, about 8.0 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (7.2 GB file, about 9.7 GB in use).
  • 16 GB VRAM: the best quant that fits is BF16 (13.4 GB file, about 15.9 GB in use).

Which granite-4.0-tiny-preview quant should you choose?

Q4_K_M is the usual starting point for granite-4.0-tiny-preview: it balances file size and quality well. Here it is a 4.2 GB file.

The Q8_0 file is 7.1 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 TQ1_0 at 1.5 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegranitehybrid
Parameters6.67B (approximate, mixture of experts)
Layers40
Trained context1,048,576 tokens
KV cache (fp16)240 KB per token

Source

Quantized byMungert
FamilyGranite
Licenseapache-2.0
Downloads (HF)734
Last updated2025-09-24

Frequently asked questions about granite-4.0-tiny-preview

How much VRAM does granite-4.0-tiny-preview need?

About 6.7 GB at Q4_K_M with an 8K context window. That is the 4.2 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 granite-4.0-tiny-preview on an 8 GB GPU?

Yes, with Q6_K_M: the 5.5 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 granite-4.0-tiny-preview GGUF?

It depends on the quant: TQ1_0 is 1.5 GB, TQ2_0 is 1.8 GB, IQ2_XS is 2.1 GB, IQ2_S is 2.2 GB, IQ2_M is 2.3 GB, Q2_K_M is 2.5 GB and 21 more.

What context length does granite-4.0-tiny-preview support?

The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 2.0 GB at 8K and 8.1 GB at 32K.

Where can I download granite-4.0-tiny-preview GGUF?

The files are on Hugging Face in the Mungert/granite-4.0-tiny-preview-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 granite-4.0-tiny-preview 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 Granite-4.0-Tiny-Preview on 8 GB VRAM?

Yes, using TQ1_0 or TQ2_0 quantizations which require about 4–4.3 GB memory at 8k context.

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.

Other GGUFs of granite-4.0-tiny-base-preview

Similar models

Browse all →

More from Mungert

All by Mungert →

More Granite models

All Granite →

Other small models

All small models →