Skip to content

Trending AI Today

XHToken/Spark-X2.6-4B

Model · Text generation · edition of October 24, 2026

XHToken/Spark-X2.6-4B is a text generation model published by XHToken on Hugging Face, a fine-tune of XHToken/Spark-X2.6-4B-Base. At the collection of October 24, 2026 it was No. 6 on the trending list, with 1,060 downloads in 30 days and 422 likes.

Collected · snapshot shipped with the site build, not a live feed

Why it is on the list

  • No. 6 by Hugging Face trending score (388)
  • repository created 5 days ago
  • +3 likes since 2026-10-23

1,060downloads in 30 days

422likes

Since October 23, 2026: +0 downloads (30-day figure), +3 likes

Positions and counts are Hugging Face’s; we report them and add no score of our own.

Open on Hugging Face (opens in a new tab)

Source: Hugging Face public API · updated

Fact sheet

Licence
Apache-2.0 · Commercial use allowedPermissive licence: commercial use is allowed, with the attribution and notice conditions of the licence.
Size
4,112,079,360 parametersAbout 8.2 GB of weights at 16 bits; about 2.3 GB at a typical 4-bit quantisation.
Formats and library
safetensors, transformers
Fine-tune of
XHToken/Spark-X2.6-4B-Base (opens in a new tab)
Access
Open download
Created · last changed
·
Downloads, all time
1,060

Most downloaded quantisations

Repositories that declare this model as base

Questions about Spark-X2.6-4B

Why is XHToken/Spark-X2.6-4B trending?

No. 6 by Hugging Face trending score (388); repository created 5 days ago; +3 likes since 2026-10-23. Hugging Face does not publish the formula of its trending score; these are the public figures behind the position.

Can Spark-X2.6-4B be used commercially?

The repository is tagged with the licence "Apache-2.0". Permissive licence: commercial use is allowed, with the attribution and notice conditions of the licence. The tag is set by the uploader, so confirm it against the licence file in the repository.

How large is Spark-X2.6-4B?

4.11B parameters according to the safetensors metadata. Weights alone take about 8.2 GB at 16 bits per parameter and about 2.3 GB in a typical 4-bit quantisation, before the memory needed for the context.

Are there quantised versions of Spark-X2.6-4B?

Yes. The most downloaded repositories that declare XHToken/Spark-X2.6-4B as their base are XHToken/Spark-X2.6-4B-GGUF (9,824 downloads in 30 days), bartowski/Spark-X2.6-4B-GGUF (1,316 downloads in 30 days), mradermacher/Spark-X2.6-4B-i1-GGUF (957 downloads in 30 days).