arXiv:2508.10925 — Paper

Specification

No structured attributes yet.

Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →

Provenance

Attributed facts

0

Source tiers

—

Freshest observation

—

Conflicts

None

gpt-oss-120b OpenAIActiveOpen weights
Most powerful open-weight model, fits into an H100 GPU
params 116.8Bcontext 131.1Kreleased 4 Aug 2025license apache-2.0
79
gpt-oss-20b OpenAIActiveOpen weights
Medium-sized open-weight model for low latency
params 20.9Bcontext 131.1Kreleased 4 Aug 2025license apache-2.0
79
gpt-oss-safeguard-20b OpenAIActiveOpen weights
gpt-oss-safeguard-20b is a safety reasoning model from OpenAI built upon gpt-oss-20b. This open-weight, 21B-parameter Mixture-of-Experts (MoE) model offers lower latency for safety tasks like content classification, LLM filtering, and trust...
params 21.5Bcontext 131.1Kreleased 18 Sept 2025license apache-2.0
70

Describes: gpt-oss-120b, gpt-oss-20b, gpt-oss-safeguard-20b

As of

Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.

Claim history

No claims recorded yet

Claims appear when a source states a fact; every later change is kept as a new claim.

11 Sept 2026
New paperPaperarXiv:2508.10925
New paper: arXiv:2508.10925
huggingface
huggingface.co

Full timeline →

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/openai/gpt-oss-safeguard-20b	model_page	T2· Quality secondary	1 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/openai/gpt-oss-120b	model_page	T2· Quality secondary	1 h ago	2
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/openai/gpt-oss-20b	model_page	T2· Quality secondary	1 h ago	2

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.