Spark-X2.5-4B-MLX-8bit safetensors model info | Hugging Bay

Q: What is this model?

abenzerps/Spark-X2.5-4B-MLX-8bit is an open-weights language model for text generation published by abenzerps, added to Hugging Bay Sep 21, 2026. Model from Hugging Face for text generation: abenzerps/Spark-X2.5-4B-MLX-8bit · mlx

Q: Can I use it commercially?

It is released under the apache-2.0 license. Commercial use is allowed. Permissive open license: commercial use, modification, and redistribution are permitted.

Q: How big is it?

abenzerps/Spark-X2.5-4B-MLX-8bit has 4B parameters. Hosted here: 6 files, 4.38 GB. Full repo upstream: 4.38 GB. The recommended download, model.safetensors, is 4.37 GB.

Q: What can I download here?

Hosted here: 6 files, 4.38 GB. Full repo upstream: 4.38 GB. The original release is at https://huggingface.co/abenzerps/Spark-X2.5-4B-MLX-8bit.

Q: File pattern scans

Automated file-pattern scans flagged nothing; not a safety guarantee. 6 hosted files carry a published SHA-256 you can re-hash. The machine-readable verdict is at /api/artifacts/hf-model-abenzerps-spark-x2-5-4b-mlx-8bit/safety on Hugging Bay.

Q: Will abenzerps/Spark-X2.5-4B-MLX-8bit run on my computer?

We can’t check the fit; we don’t know your machine. We don’t have your machine’s specs, so no check ran and no verdict is shown. Answer two quick questions and we’ll estimate, from the memory you tell us, whether it fits your machine. Or skip it; you can still download; fit isn’t checked.

abenzerps

Spark X2.5 4B MLX 8bit

Code

This receipt records checks for this exact file; it does not prove safety or that it will run on your device. Read the receipt

Model details

text generationLicense: apache-2.0SHA-256 recorded for 6 of 14 filessafetensors

Can I run it?View as JSON →

Quick answer: compatibility and sources

Can I run Spark-X2.5-4B-MLX-8bit on my computer?

It depends on your memory, processor, file format and runtime. A download is not confirmed by the current file evidence; inspect the file record before trying it. A file-size fit is an estimate, not a successful run or speed promise. Read the license before use.

Spark-X2.5-4B-MLX-8bit is an open-weights language model for text generation. Model weights are learned parameters; a runtime loads them. A hosted file is an individual file Hugging Bay can serve. SHA-256 identifies exact file bytes. The availability date above applies to this file record, not the upstream metadata update date.

As of the recorded file availability check on 2026-10-29.

Recorded SHA-256 for model.safetensors: 2471dc02ed61ba53907a12e8d941a7d13e5e1e17cf16f9354e8487bb7e105149. Exact file evidence · Machine-readable source facts · Run evidence and any recorded receipts. A checksum fingerprints only the named file bytes; it does not prove safety, quality, a complete multi-file selection or a successful run.

Updated 2026-10-21 · Page-copy source revision 8871bab84e1cf80aea1686a944ec335078a6c036 · Source files: apps/web/lib/model-page-template.mjs, apps/web/lib/model-page-v3.mjs, apps/web/lib/aeo-page.mjs · Full revision record

Model metadata: As of 2026-09-21 at the model source. Popularity counters are source-reported, not measured by Hugging Bay. Availability checked 2026-10-29.

Key facts

Catalog record: Summarizes upstream metadata; it is not itself a download or run receipt.
File evidence: Lists exact filenames, availability and any recorded SHA-256 for individual files.
Run evidence: Reports measured results only when a receipt exists for named file bytes and a setup; otherwise unmeasured.
Checksum scope: SHA-256 fingerprints one file’s bytes; it does not prove safety, quality or a multi-file bundle.

Evidence you can inspect (plain text)

This page’s quick answer names the file record for hf-model-abenzerps-spark-x2-5-4b-mlx-8bit.
Open /artifact/hf-model-abenzerps-spark-x2-5-4b-mlx-8bit/files for this exact filenames, availability and any recorded SHA-256.
Open /evidence/artifacts/hf-model-abenzerps-spark-x2-5-4b-mlx-8bit for run-readiness evidence tied to that record; absent receipts stay unmeasured.
Open /api/artifacts/hf-model-abenzerps-spark-x2-5-4b-mlx-8bit/card.md for machine-readable source facts (not a runtime test).
A matching checksum on this page fingerprints only the named bytes shown in file evidence.

Unmeasured states are stated explicitly; this page does not invent run scores or compatibility guarantees.

Which date applies to which fact

File availability check: 2026-10-29
Committed page-copy sources: 2026-10-21
Upstream model metadata: Dates on artifact records when shown; not implied by page-copy revision
Run receipts: Only on /evidence pages when recorded; otherwise unmeasured

Page-copy revision (Updated line above) identifies committed text sources, not upstream model edits or test outcomes.

Requirements and source record

Try browser chat with a small demo model (a separate model; device support is checked before loading).

abenzerps/Spark-X2.5-4B-MLX-8bit

Content controls: not assessed.

The catalog lists this model for having conversations. Catalog sources: license, cardMetadata.license, tags, task, cardMetadata.pipelineTag.

Download availability is being checked.

Hashes recorded. Date not recorded.

File	Size	SHA-256	Source	Download
Weights
`model.safetensors`	4.37 GB	`2471dc02…7e105149` `2471dc02ed61ba53907a12e8d941a7d13e5e1e17cf16f9354e8487bb7e105149`	Download availability is being checked.	File metadata
Tokenizer & config
`SHA256SUMS.txt`	84 bytes	—	Download availability is being checked.	File metadata
`SPARK_MLX_LLM_LICENSE`	11.4 KB	—	Download availability is being checked.	File metadata
`assets/benchmark.png`	64.1 KB	—	Download availability is being checked.	File metadata
`chat_template.jinja`	4.65 KB	`70534572…ee77a9ba` `705345721970da40e6aa56db548a46ecd1d8d94a5418808fa180ac19ee77a9ba`	Download availability is being checked.	File metadata
`config.json`	2.5 KB	`4d0b03bb…6ca7686a` `4d0b03bb66fb79e1774ac233eb918734a42471be5ca048f09affd69f6ca7686a`	Download availability is being checked.	File metadata
`configuration_spark.py`	4.33 KB	—	Download availability is being checked.	File metadata
`generation_config.json`	281 bytes	`804a5276…32ad9e6c` `804a5276a9d237600688267dd2bf6f9d4129ec21faf417f870d5d52b32ad9e6c`	Download availability is being checked.	File metadata
`model.safetensors.index.json`	51.7 KB	—	Download availability is being checked.	File metadata
`modeling_spark.py`	19.4 KB	—	Download availability is being checked.	File metadata
`tokenizer.json`	10.1 MB	`f71f91ee…02454c49` `f71f91ee1b2ac0b27e47bf49ee4859d322ab3d191217ec9964d0189b02454c49`	Download availability is being checked.	File metadata
`tokenizer_config.json`	389 bytes	`77cccc98…45032b85` `77cccc9891b2b93af38bd2010edc8271568f80b53c74ea6fc1b6729945032b85`	Download availability is being checked.	File metadata
Docs
`.gitattributes`	1.52 KB	—	Download availability is being checked.	File metadata
`README.md`	1.37 KB	—	Download availability is being checked.	File metadata

Check a download

In the folder where you saved chat_template.jinja, run:

macOS

shasum -a 256 -- 'chat_template.jinja'

Linux

sha256sum -- 'chat_template.jinja'

Windows

Get-FileHash -LiteralPath 'chat_template.jinja' -Algorithm SHA256

Compare the result with the recorded SHA-256.