CatalogWantedMirrorsLedgerOperating StatementDocsStudio statusCard system
alt.freeweights.gpt-oss / gpt-oss-safeguard-20b
From: openai <huggingface.co/openai>Newsgroups: alt.freeweights.gpt-ossSubject:

gpt-oss-safeguard-20b

Date: 2026-11-14Commit: 8a11e17Signed: 2DC28CC7B26A283B ✓

Text generation model with 21.5B parameters. Safetensors weights, 13.8 GB in all. Published by openai under Apache-2.0.

21.5B params13.8GBgpt-oss-safeguard-20bsafetensorspermissive
FactsParams21.5BSize13.8 GBFormatssafetensorsTasktext generationLicenseApache-2.0From sourcepermissive
$ fw get openai/gpt-oss-safeguard-20b
# manifest sha256 694538dc6f54… · 13.8 GB · 15 files
$ fw verify ./gpt-oss-safeguard-20b openai/gpt-oss-safeguard-20b
From sourceHosted (signed)openai/gpt-oss-safeguard-20bLast check: 2026-11-14 21:24 UTC
Run at homeGet help

Get help: custom fine-tune or private deploy. We review your setup before we quote.

Fine-tune of gpt-oss-20b.

Works with

Run24Fine-tune29Convert15Evaluate13Scan13Apps58
OverviewModel cardFilesUse itLicenseProvenanceModel tree

Overview

Size
13.8 GB
Weights
13.8 GB (safetensors)
Parameters
21.5B
Formats
safetensors
Task
text generation
License
Apache-2.0 (permissive)
Files
15, each with a sha256
Publisher
openai
Added
2026-11-14
Section
alt.freeweights.gpt-oss
Lineage

Original release from the publisher.

Base model: openai/gpt-oss-20b

Where these bytes come from
Publisher at the pinned commit→You

Sources are tried in order. A file counts only when its size and sha256 match the manifest. Sources are checked when you ask for them, not on a timer.

Sources
BadgeHostLast checkResult
From sourcehuggingface.co2026-11-14 21:24 UTCpassed

Model card

From the publisher at commit 8a11e17. Shown as plain text with links; images open at the publisher.

Image: gpt-oss-safeguard-20b

Try gpt-oss-safeguard · Guide · Model card · OpenAI blog

gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases. For other applications, we recommend using gpt-oss models.

This model gpt-oss-safeguard-20b (21B parameters with 3.6B active parameters) fits into GPUs with 16GB of VRAM. Check out gpt-oss-safeguard-120b (117B parameters with 5.1B active parameters) for the larger model.

Both models were trained on our harmony response format and should only be used with the harmony format as it will not work correctly otherwise.

Highlights#

  • Trained to reason about safety : Trained and tuned for safety reasoning to accommodate use cases like LLM input-output filtering, online content labeling and offline labeling for Trust and Safety use cases.
  • Bring your own policy: Interprets your written policy, so it generalizes across products and use cases with minimal engineering.
  • Reasoned decisions, not just scores: Gain complete access to the model’s reasoning process, facilitating easier debugging and increased trust in policy decisions. Keep in mind Raw CoT is meant for developers and safety practitioners. It’s not intended for exposure to general users or use cases outside of safety contexts.
  • Configurable reasoning effort: Easily adjust the reasoning effort (low, medium, high) based on your specific use case and latency needs.
  • Permissive Apache 2.0 license: Build freely without copyleft restrictions or patent risk—ideal for experimentation, customization, and commercial deployment.

Inference examples#

You can use gpt-oss-safeguard-120b and gpt-oss-safeguard-20b similar to gpt-oss-120b and gpt-oss-20b as described in our respective cookbooks. We’ve also provided a detailed prompting guide that provides guidelines for how to craft your policy and use it with the models.

Download the model#

To download the model weights from Hugging Face hub using similar instructions to gpt-oss-120b.

Join the ROOST Model Community#

gpt-oss-safeguard is a model partner of the Robust Open Online Safety Tools (ROOST) Model Community. The ROOST Model Community (RMC) is a group of safety practitioners exploring open source AI models to protect online spaces. As an RMC model partner, OpenAI is committed to incorporating user feedback and jointly iterating on future releases in pursuit of open safety. Visit the RMC GitHub repo to learn more about this partnership and how to get involved.

Resources#

Files

15 files · 13.8 GBWeights 13.8 GB (safetensors). Sizes and sha256 from the signed manifest.
Hide all 15 files with their sha256
FileSizesha256
.gitattributes1.57 KB34448b82c17d60fec9b65b1f093c115ddbaadc04beb1b0140b6bfed2e012a930
LICENSE11.4 KBcfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30
README.md4.03 KB1a3a493bbb99dd2935e6e61c902be60ba326d7627958f5f78279259bb77027e9
USAGE_POLICY236 Ba48000373022ca5114dc538e01085a3885884f41d4ba5c12dcd30d1e2d21dadf
chat_template.jinja16.7 KB589c73db2f7775c5f946b84bd4695fd3be3612e48ebb69a956dc38b7ec70236d
chat_template.json17.5 KBc8f5c2924c29dc0c6fafb1894863d620b2c0b065a8aef860c79f65b911ef95a7
config.json1.81 KB8fc6c94451a1a97bf3080baec34002c75173833164bd8e75f40152e27b31e531
generation_config.json165 B97b165839e19bf43309d1e571d1720ebc081dde24e6722cfb644257215ad1e66
model-00000-of-00002.safetensors4.79 GBef82af3aff019605840ebd545957b8493a65a179a8c11339e50753d311195d12
model-00001-of-00002.safetensors4.80 GB19684279c3bf220c0bf6e63a86ee22a2724debb6972c7ad59b0d8aa42dff642d
model-00002-of-00002.safetensors4.17 GBe18569972d3995dd7327e4423a6a3de5b1356223d4f710808bfcbac09f2b4d5d
model.safetensors.index.json36.4 KB0e085b977c4c9942f85938828e8c989ed7d5cdabf852e4da6a67c116cd502cd1
special_tokens_map.json98 Bdd5e191d20c12d2fee1da5bae14ca1db0f5f4215300af691f23cdee97120a293
tokenizer.json27.9 MB0614fe83cadab421296e664e1f48f4261fa8fef6e03e63bb75c20f38e37d07d3
tokenizer_config.json4.20 KB9279e942392b742d633c7adbb89ebe002c98399db8926a7af5125c726f404070
VersionsEach time the pinned upstream commit or file list changed in this repo. Newest first.
Upstream commitDateContents
8a11e17b25 pinned now manifest updated, same commit2026-11-14 21:24 UTC15 files · 13.8 GB
8a11e17b252026-11-14 21:24 UTC15 files · 13.8 GB

Use it

$ fw get openai/gpt-oss-safeguard-20b
# manifest sha256 694538dc6f54… · 13.8 GB · 15 files
$ fw verify ./gpt-oss-safeguard-20b openai/gpt-oss-safeguard-20b
fw.pyChecks the signature and every sha256, tries each source in order, resumes.
$ python3 fw.py get https://freeweights.allternit.com/manifests/openai/gpt-oss-safeguard-20b.json --dir ./gpt-oss-safeguard-20b $ python3 fw.py verify ./gpt-oss-safeguard-20b ./openai__gpt-oss-safeguard-20b.json
hfThe Hugging Face CLI with one setting. Same repo id.
$ HF_ENDPOINT=https://freeweights.allternit.com hf download openai/gpt-oss-safeguard-20b --revision 8a11e17b25c973a24099d4016bf2e17dd7ec1574 --local-dir ./gpt-oss-safeguard-20b
transformersLoad the weights in Python.
import os os.environ["HF_ENDPOINT"] = "https://freeweights.allternit.com" from transformers import AutoModelForCausalLM, AutoTokenizer tok = AutoTokenizer.from_pretrained("openai/gpt-oss-safeguard-20b", revision="8a11e17b25c973a24099d4016bf2e17dd7ec1574") model = AutoModelForCausalLM.from_pretrained("openai/gpt-oss-safeguard-20b", revision="8a11e17b25c973a24099d4016bf2e17dd7ec1574", device_map="auto")
vLLMServe an OpenAI-compatible API.
$ HF_ENDPOINT=https://freeweights.allternit.com vllm serve openai/gpt-oss-safeguard-20b --revision 8a11e17b25c973a24099d4016bf2e17dd7ec1574
Run at home

License

Name
Apache License 2.0
Id
Apache-2.0
Class
Permissive
Redistribution
yes
Listing
Hosted (signed)
Decided
License as declared by the publisher. No person read it. 2026-11-14
Text in
LICENSE
Text sha256
cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30
Terms gate
No extra terms beyond the license.

Read the license text

Licenses are checked against the license table. A custom license passes only after a person reads it.

Provenance

From: openai <huggingface.co/openai>Newsgroups: alt.freeweights.gpt-ossSubject: gpt-oss-safeguard-20bDate: 2026-11-14Commit: 8a11e17Signed: 2DC28CC7B26A283B ✓
Upstream
openai/gpt-oss-safeguard-20b
Pinned commit
8a11e17b25c973a24099d4016bf2e17dd7ec1574
Manifest
gpt-oss-safeguard-20b.json · .minisig
Manifest sha256
694538dc6f54c6807e521d854a97a883e87db7cd6d6057f4e04d8c788f46b755
Signature
minisign, Ed25519, key 2DC28CC7B26A283B
Created
2026-11-14 21:24 UTC
Generator
freeweights-manifest/0.1.0
Takedowns
None.

How to verify

$ curl -O https://freeweights.allternit.com/manifests/openai/gpt-oss-safeguard-20b.json $ curl -O https://freeweights.allternit.com/manifests/openai/gpt-oss-safeguard-20b.json.minisig $ minisign -Vm gpt-oss-safeguard-20b.json -P RWQ7KGqyx4zCLXtUFOqvfv8L3g+8fpKJ1FsuxPyOYMF+4KufbUxKq1Oq $ shasum -a 256 <file> # compare with the Files tab

A manifest whose signature fails must be rejected. A file whose size or sha256 does not match must be discarded. Verify a download

Model tree

Parent

openai/gpt-oss-20b · this model is a finetune of it

3 models in the catalog name this one as their parent.

Quantizations
3
Quantizations 3
Get it
Download
$ HF_ENDPOINT=https://freeweights.allternit.com hf download openai/gpt-oss-safeguard-20b
Open in Freeweights Studio

Pull, verify and run it in the Allternit app on Desktop, web or phone.

Signed manifest
minisign, key 2DC28CC7B26A283B. gpt-oss-safeguard-20b.json. Provenance. Takedown history: none.
Source
huggingface.co/openai/gpt-oss-safeguard-20b

Variants

3 quantized or adapter builds of gpt-oss-safeguard-20b. Each has its own license check and page.

SubjectSizeParamsLicenseDate
gpt-oss-safeguard-20b-MLX-MXFP4Hosted (signed) From: lmstudio-community in alt.freeweights.gpt-oss11.2 GB20.9BPermissive2026-11-15gpt-oss-safeguard-20b-GGUFHosted (signed) From: lmstudio-community in alt.freeweights.gpt-oss12.1 GB—Permissive2026-11-15gpt-oss-safeguard-20b-GGUFHosted (signed) From: unsloth in alt.freeweights.gpt-oss191 GB—Permissive2026-11-15

Related models

All of alt.freeweights.gpt-oss →
From: unsloth <huggingface.co/unsloth>Subject: gpt-oss-120b
From: mlx-community <huggingface.co/mlx-community>Subject: gpt-oss-120b-4bit
From: unsloth <huggingface.co/unsloth>Subject: gpt-oss-120b-BF16
From: lmstudio-community <huggingface.co/lmstudio-community>Subject: gpt-oss-120b-GGUF
From: unsloth <huggingface.co/unsloth>Subject: gpt-oss-120b-GGUF
From: lmstudio-community <huggingface.co/lmstudio-community>Subject: gpt-oss-120b-MLX-8bit