openai/gpt-oss-20b — OpenHubAI

In plain English

Good for triaging log lines and alerts into plain English summaries so a person can prioritise faster. Not a substitute for a real detection pipeline, and it should never auto close an alert on its own.

OpenAI's 20 billion parameter open weight model, tuned for reasoning style tasks and released with permissive licensing for local deployment.

Required documentation

Complete

Intended use

Intended for summarising and triaging text such as logs or alerts for a human operator. Not intended to make automated security decisions.

Training data provenance

OpenAI's open weight reasoning model, released with its own model card describing training data composition at a high level.

Licence & terms

Apache-2.0.

Reviews3

Write a review

MRMaya Rehanirep 4.8Production6cee5e8

7h ago

Accuracy4.1

Fine-tune3.5

Documentation3.7

Speed3.4

TNTomasz Nowakrep 4.5Prototype6cee5e8

7h ago

Accuracy3.9

Fine-tune3.6

Documentation3.9

Speed3.2

PSPriya Shahrep 4.6Production6cee5e8

7h ago

Accuracy4.2

Fine-tune3.4

Documentation3.6

Speed3.5

Discussion0

Posting as a signed-in userSign in to post

No questions yet. Be the first to ask one.

Structured rating

3.7

from 3 reviews of 6cee5e8

Only 3 reviews on 6cee5e8 so far. Earlier versions carry more, but they are not this build.

Accuracy4.1

Fine-tune3.5

Documentation3.7

Speed3.4

Run it now

ollama run openai/gpt-oss-20b

Will it fit?

Build	File	VRAM
Q4_K_M	12.0 GB	14.0 GB
Q8_0	21.5 GB	23.5 GB

Set a hardware profile in the browse sidebar to see which builds fit.

Inference providers

No hosted providers listed yet.

No cloud required. Run it with

vLLMSGLang

Model tree

Base modelNone, this is a base model

Adapters0

Finetunes0

Quantizations0

Merges0

At a glance

Parameters—

Licenceapache-2.0

Downloads6.6M

Reviews3

Docs completenessComplete