GPT-OSS-120B: gPQA Diamond now reports OpenAI’s published launch result of 80.1%; results from different evaluation settings are not interchangeable

The OnAirToday AI Progress Tracker recorded a verified change to OpenAI's GPT-OSS-120B (version gpt-oss-120b): GPQA Diamond now reports OpenAI’s published launch result of 80.1%; results from different evaluation settings are not interchangeable.

GPT-OSS-120B is OpenAI’s open-weight, text-based reasoning model, released under the Apache 2.0 license for customizable and self-hosted deployments. It supports configurable reasoning effort and tool use, with documented deployment on a single 80 GB GPU, but requires application-level safeguards and factual verification.

What changed in this review:

• SWE-bench Verified is included at 62.4%, using OpenAI’s published launch evaluation.
• Token pricing depends on the hosting provider and deployment; no universal model-wide input or output rate applies.
• OpenAI subsequently released gpt-oss-safeguard variants for safety classification; these are distinct from the base gpt-oss-120b reasoning model.
• Flaws and limitations updated
• Risks and threats updated

Teams building on GPT-OSS-120B should confirm limits, pricing and benchmark coverage against the vendor's own documentation before relying on these figures.

The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/gpt-oss-120b.

References:

• https://openai.com/index/introducing-gpt-oss/
• https://huggingface.co/openai/gpt-oss-120b
• https://openai.com/index/gpt-oss-model-card/
• https://openai.com/index/introducing-gpt-oss-safeguard/